Reaction optimisation

Abstract

The present invention relates to a method of optimising an instruction set for a chemical reaction, a chemical synthesiser for use in the method, and a chemical synthesis platform comprising the chemical synthesiser. The method includes (i) obtaining an initial instruction set for a chemical reaction, (ii) generating derivative instruction sets from the initial instruction set using a predictive model, where each derivative instruction set is different, (iii) performing chemical reactions according to the derivative instruction sets, where each chemical reaction is performed in a different reaction vessel of a chemical synthesiser, (iv) recording reaction data during and / or product data using an analytical unit of the chemical synthesiser, (v) updating the predictive model using the recorded reaction data and / or product data to optimise a characteristic of the chemical reaction; and (vi) generating an optimised instruction set based on the updated predictive model.

Description

[0001] REACTION OPTIMISATION

[0002] Related Application

[0003] The present application claims the benefit of, and priority to, GB 2502560.2 filed on 21 February 2025 (21.02.2025), the contents of which are hereby incorporated by reference in their entirety.

[0004] Field of the Invention

[0005] The present invention relates to a method of optimising an instruction set for a chemical reaction, a chemical synthesiser for use in a method of optimising an instruction set for a chemical reaction, and a chemical synthesis platform including the chemical synthesiser.

[0006] Background

[0007] Methods and systems for performing automated chemical synthesis involve the performance of chemical reactions, synthesis and discovery of new compounds, and the optimisation of their reaction conditions and reaction products, typically without any direct involvement from a human user. Such methods and systems aim to improve the speed, precision, and reproducibility of chemical reactions, and reduce the need for an experienced chemist to carry out repetitive synthetic procedures, which ultimately advances the capabilities of chemical research and development.

[0008] Chemical synthesis requires intensive, highly skilled labour and a typical laboratory-scale synthesis can require multiple complex synthetic operations that are difficult to explicitly encode. By using machine-readable instruction sets developed from digital models that are responsive to the machine-readable instructions, it is possible to develop and utilise standardised models for chemical synthesis that can be executed by automated synthesisers. Such machine-readable instruction sets are commonly adapted from chemical literature and follow a strict, predetermined synthetic operation.

[0009] Chemical synthesis optimisation traditionally focuses on an iterative optimisation process, which is intended to involve the gradual improvement of a chemical synthesis through repeated cycles of reaction, analysis and adjustment to reaction conditions. The iterative adjustment after each individual reaction step may be used to explore a reaction space or to optimise a reaction.

[0010] However, iterative optimisation has limitations. Refinement of reaction parameters over a long series of chemical reactions can be time-consuming and resource intensive. For example, an operator may need to load reactants and reset the reaction for each iteration in the series. Furthermore, the approach may inherently lead to a convergence towards a local

[0011] 008889958optimum, as data on the reaction is fed into a prediction model at the end of each reaction in the series.

[0012] There is a need for improved approaches to optimising chemical reactions.

[0013] Summary of the Invention

[0014] The present inventors have devised a method and chemical synthesiser for the optimization of chemical reactions which look to address the above-mentioned problems. An initial instruction set for a chemical reaction is provided, and is used to generate a number of derivative instruction sets. The derivative instructions sets (and optionally the initial instruction set) are performed in different reaction vessels of a chemical synthesiser.

[0015] Reaction data and product data on the reactions are used for process state monitoring and to quantify reaction outcomes, and is used to update a predictive model for the optimisation of instruction sets.

[0016] Generally, the invention relates to a method of optimising a chemical reaction. The method may be performed autonomously by a chemical synthesiser.

[0017] In general, the present invention provides a method of optimising an instruction set for a chemical reaction, the method comprising the steps of:

[0018] generating a number of derivative instruction sets from an initial instruction set using a predictive model, where each of the derivative instruction sets is different;

[0019] performing chemical reactions according to the number of derivative instruction sets and optionally the initial instruction set;

[0020] recording reaction data during each of the chemical reactions and / or recording product data on a reaction product of each of the chemical reactions; and

[0021] updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction.

[0022] In a general aspect, the present invention provides a method of optimising a chemical synthesis, the optimisation comprising the steps of:

[0023] (i) obtaining an initial instruction set for a chemical reaction,

[0024] (ii) generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different,

[0025] (iii) performing chemical reactions according to the number of initial instruction set and / or each of the derivative instruction sets, wherein each of the chemical reactions is performed in a chemical synthesiser; and

[0026] (iv) recording reaction data using an analytical unit of the chemical synthesiser during each of the chemical reactions and / or recording product data on a reaction product of each of the chemical reactions;

[0027] 008889958(v) updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction; and

[0028] (vi) generating an optimised instruction set based on the predictive model.

[0029] In a first aspect of the invention there is provided a method of optimising an instruction set for a chemical reaction, the method comprising the steps of:

[0030] (i) obtaining an initial instruction set for a chemical reaction;

[0031] (ii) generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different;

[0032] (iii) performing chemical reactions according to the number of derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performed in a different reaction vessel of a chemical synthesiser;

[0033] (iv) recording reaction data during each of the chemical reactions and / or recording product data on a reaction product of each of the chemical reactions using an analytical unit of the chemical synthesiser;

[0034] (v) updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction; and

[0035] (vi) generating an optimised instruction set based on the updated predictive model.

[0036] The present invention provides for a method which makes use of a particular optimisation approach, whereby each of the chemical reactions is performed in a different reaction vessel. In order words, a chemical reaction performed in a first reaction vessel is independent of a chemical reaction performed in a second vessel. This may allow for each of the chemical reactions to be performed in a parallel manner, such that each of the chemical reactions may be performed concurrently.

[0037] Advantageously, the method may also improve the time and resources efficiency of the optimisation procedure. By performing reactions in parallel, a number of reactions may be set up and performed at the same time, requiring less operator involvement. Once the reactions are complete, the operator can also reset the platform and set up the next set of reactions in one go. This reduces the time spent by the operator per reaction performed.

[0038] Typically, more than one derivative instruction sets are generated. A plurality of derivative instruction sets may be generated. The number of derivative instruction sets generated from the initial instruction set is typically one or more, such as two or more, such as three or more, such as four or more, such as five or more, such as six or more, such as seven or more, such as eight or more, such as nine or more, such as ten or more.

[0039] The method may also provide optimisation routes to the exploration of a wider chemical space, such that at least local performance maxima, and preferably global performance

[0040] 008889958maxima, may be discoverable. By generating a number of derivative instruction sets at the start of the process, this ensures that a broader range of the chemical reaction space is explored before the predictive model is updated. This increases the chance that a global optimum is identified. This contrasts to series optimisations, which may be only identify locally optimised reaction conditions as the predictive model is updated after each iteration.

[0041] The present invention provides for a method including a broader range of chemical procedures and reaction steps, and which uses data on the reaction product and also recorded during the reaction itself to drive the optimisation of the chemical synthesis. Using an array of analytical sensors to monitor the reaction mixture, the reaction environment (e.g., headspace) as well as the surrounding conditions, provides a more detailed picture of the reaction. Leveraging the combination of product data and reaction data allows for more effective reaction optimisation, beyond that of previous manual optimisation efforts and previous automated optimisation attempts. This method may be shown to improve reaction product yield, purity of the reaction product (reduction of impurity), increase starting material conversion, and is also thought to be helpful in reducing reagent or solvent usage.

[0042] The number of derivative instruction sets to be generated may be based on the complexity of the initial instruction set. The initial instruction set is analysed by the predictive model. The predictive model determines the number of derivative instruction sets and generates all the necessary derivative instruction sets for the chemical reaction.

[0043] Therefore, a user can know the number of chemical reactions needed for optimisation in advance of initiating and performing any chemical reactions, so that each of the chemical reactions can be performed and the reaction data and / or product data may be recorded for each of the chemical reactions. This may also result in reduced time and resource costs involved in the optimisation.

[0044] For example, for a low complexity reaction the number of derivative instruction sets to be generated will be relatively small - as the optimisation is expected to be relatively straightforward. However, for a high complexity reaction the number of derivative instruction sets to be generated will be relatively large - as the optimisation is expected to be more difficult. This step avoids wasting time and resource optimising straightforward reactions, and allows for more time and resource to be spent on the more difficult reactions. This complexity prediction step is particularly beneficial for parallel reaction optimisations, where there is no feedback on the reaction until after a set of reactions has been carried out.

[0045] Thus, in some embodiments, the method further comprises determining the complexity of the initial instruction set using the predictive model, and calculating the number of derivative instruction sets to be generated based on the complexity of the initial instruction set.

[0046] 008889958In some embodiments, the method further comprises determining the complexity of the chemical reaction using the predictive model, and calculating the number of derivative instruction sets to be generated based on the complexity of the chemical reaction. This may be the case where the reaction is new, and an initial instruction set (e.g., from the literature) is not available.

[0047] In such embodiments, the method may comprise a step of performing chemical reactions according to each of the derivative instruction sets generated.

[0048] In some embodiments, the complexity of the initial instruction set is based on a probability that performing a chemical reaction according to the initial instruction set will result in one or more desired characteristics of the chemical reaction.

[0049] The desired characteristic of the chemical reaction may be a characteristic of the reaction parameter, such as a reaction condition. The desired characteristic of the chemical reaction may be a characteristic of the reaction product.

[0050] In some embodiments, the complexity of the initial instruction set is based on a prediction of the probability that performing a chemical reaction according to the initial instruction set will produce a reaction product with a desired characteristic. In some embodiments, performing a chemical reaction according to at least one derivative instruction set will result in obtaining a desired characteristic of the chemical reaction.

[0051] In some embodiments, performing a chemical reaction according to at least one derivative instruction set produces a reaction product with a desired characteristic.

[0052] In some embodiments, the complexity of the initial instruction set is based on:

[0053] (a) the degree of structural difference between the starting material and the reaction product of the chemical reaction;

[0054] (b) the reaction type of the chemical reaction;

[0055] (c) the reaction mechanism of the chemical reaction; and / or

[0056] (d) literature reports of the chemical reaction.

[0057] In some embodiments, the complexity of the initial instruction set is determined based on the structural difference between the starting material and the reaction product of the chemical reaction defining the initial instruction set. Where a significant structural difference exists between the starting material and the reaction product, the complexity of the chemical reaction may be high, and therefore the number of derivative instruction sets may be high to reflect the probability of producing a reaction with a desired characteristic. The degree of structural difference may be quantified by the number of functional groups in the starting material, the number of functional groups in the reaction product, or the difference in the number of functional groups between the starting material and the reaction product.

[0058] 008889958In some embodiments, the complexity of the initial instruction set is determined based on the reaction type of the chemical reaction. For example, the complexity of one reaction type (e.g., a click reaction) may be known to be easier to optimise than of another reaction type (e.g., an aldol reaction). Thus, the complexity of other reactions of the same type is likely to be similar.

[0059] In some embodiments, the complexity of the initial instruction set is determined based on the reaction mechanism of the chemical reaction. For example, the complexity of one reaction mechanism may be known to be easier to optimise than of another reaction mechanism. Thus, the complexity of other reactions of the same mechanism are likely to be similar. For example, the complexity may be dependent on the stability of transition states during the reaction, the steric hindrance around reaction functional groups of the starting material, and potential competing reactions.

[0060] In some embodiments, the complexity of the initial instruction set is determined based on literature reports of the chemical reaction. For example, the yield provided in literature reports may be used to determine complexity. A high average yield reported may indicate that a reaction is less complex than a low average yield. In addition or alternatively, a high variance in the average yield reported may indicate that a reaction is more complex than a low variance in the average yield.

[0061] The complexity of the initial instruction set may be based on any combination of the above factors. The factors may be weighted depending on their importance in the complexity prediction.

[0062] In some embodiments, the complexity of the initial instruction set is compared to a threshold, and where the complexity is at or above the threshold the performing step is carried out on a chemical synthesiser having multiple reaction vessels, and where the complexity is below a threshold the performing step is carried out on a chemical synthesiser with a single reaction vessel.

[0063] The number of derivative instruction sets to be generated may also be based on the availability of the chemical synthesiser for performing the optimisation method. The availability of the chemical synthesiser may refer to one or more constraints which may be associated with the chemical synthesiser, such as the number of available reaction vessels of the chemical synthesiser, the availability of reagents or starting materials required to perform the chemical reactions, and the operational constraints of the chemical synthesiser. The number of instruction sets may therefore be constrained by limitations of the hardware at the time an optimisation campaign is needed.

[0064] For example, where the chemical synthesiser has reduced availability, the number of derivative instruction sets to be generated will be relatively small. However, where the

[0065] 008889958chemical synthesiser has greater availability, the number of derivative instruction sets to be generated will be relatively large. This ensures that the chemical reactions to be performed based on the derivation instruction sets are guaranteed to be executable and performable on the available chemical synthesiser, without requiring operator intervention. This step is particularly beneficial for parallel reaction optimisations, where the chemical synthesiser may be required to perform multiple chemical reactions simultaneously, and autonomously.

[0066] Where a number of derivative instruction sets is initially determined by the complexity of the initial instruction sets, that number of derivative instruction sets may be cross-checked with the availability of the chemical synthesiser for performing the optimisation method.

[0067] In some embodiments, the method further comprises determining the availability of the chemical synthesiser.

[0068] In some embodiments, the availability of the chemical synthesiser is based on:

[0069] (a) the number of available reaction vessels of the chemical synthesiser;

[0070] (b) the amount of available starting material;

[0071] (c) the operational status of the chemical synthesiser; and / or

[0072] (d) limitations defined by an operator

[0073] Preferably, the number of derivative instruction sets to be generated may be based on the complexity of the initial instruction set and the availability of the chemical synthesiser. In this way, the number of derivative instruction sets to be generated may provide a balance between reaction space exploration and time and resource efficiency, where too many derivative instruction sets may consume excess time and resource, and too few derivative instruction sets may not fully explore the reaction space.

[0074] In some embodiments, the number of derivative instruction sets to be generated is determined by first determining the availability of the chemical synthesiser, then determining the complexity of the initial instruction set.

[0075] In this way, the availability of the chemical synthesiser may serve to provide an indication of a maximum number of derivative instruction sets to be generated, and thus a maximum number of chemical reactions to be performed.

[0076] In other embodiments, the number of derivative instruction sets to be generated is determined by first determining the complexity of the initial instruction set, then determining the availability of the chemical synthesiser.

[0077] In this way, the complexity of the initial instruction set may serve to provide an indication of a maximum number of derivative instruction sets to be generated, and thus a maximum number of chemical reactions to be performed.

[0078] 008889958In such embodiments, the chemical synthesiser may be part of a chemical synthesis platform. The chemical synthesis platform may include a chemical synthesiser of the invention having a plurality of reaction vessels and a different chemical synthesiser having only one reaction vessel. In this way, when the reaction is more complex (above the threshold), the synthesiser with a plurality of reaction vessels is used in order to efficiently optimise the reaction, but when the reaction is less complex (below the threshold), the synthesiser with a single reaction vessel is used. Thus, the chemical synthesiser are used efficiently, with parallel optimisation only be used for more complex reactions.

[0079] Steps (ii) to (v) may be repeated one or more times. The number of times steps (ii) to (v) are repeated may depend on the complexity of the reaction (as discussed above), or may depend on if the reaction has been sufficiently optimised. Thus, in some embodiments, steps (ii) to (v) are repeated based on the updated predictive model, wherein step (ii) comprises generating a number of derivative instruction sets from the updated predictive model.

[0080] In some embodiments, steps (ii) to (v) are repeated until the one or more characteristics of the chemical reaction are optimised. The chemical reaction may be considered optimised when the one or more characteristics of the reaction have reached a threshold level. In addition or alternatively, the chemical reaction may be considered optimised when improvements in the one or more characteristics of the reaction have plateaued from successive repeats.

[0081] After steps (ii) to (v) have been repeated, then step (vi) may then be generated based on the updated predictive model.

[0082] In more detail, the method may comprise:

[0083] (ii-a) generating a number of derivative instruction sets from the updated predictive model, where each of the derivative instruction sets is different;

[0084] (iii-a) performing chemical reactions according to each of the derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performed in a different reaction vessel of a chemical synthesiser to produce a reaction product for each of the chemical reactions;

[0085] (iv-a) recording reaction data during each of the chemical reactions and / or recording product data on the reaction product of each of the chemical reactions using the analytical unit of the chemical synthesiser;

[0086] (v-a) updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction; and

[0087] (vi-a) repeating steps (ii-a) to (v-a) until the one or more characteristics of the chemical reaction are optimised.

[0088] 008889958It is an aim of the method of the present invention to optimise synthetic routes by optimising a characteristic of a chemical reaction. Upon recording reaction data and / or product data, this is done by adjusting one or more reaction parameters of an instruction set, such as the initial instruction set or one of the derivative instruction sets.

[0089] For example, the optimisation of a characteristic of a chemical reaction may be the optimisation of producing synthetic routes to reaction products having a desired characteristic. Such desired characteristics include improved reaction product yield, improved purity of the reaction product and increase in starting material conversion.

[0090] In some embodiments, the one or more characteristics of the chemical reaction is reaction product yield, reaction product purity, reaction product to starting material ratio, reaction product to impurity ratio, solvent mass consumption, reagent mass consumption, reaction duration or a combination thereof. Preferably, the one or more characteristics is reaction product yield, and the yield is determined by NMR spectroscopy, Raman spectroscopy or high-performance liquid chromatography product data.

[0091] In some embodiments, each of the derivative instruction sets is different to the initial instruction set. Preferably, a range of derivative instruction sets is produced so that a wider range of reaction parameters may be covered by the method.

[0092] In some embodiments, each of the chemical reactions is performed before the predictive model is updated for optimisation. In particular, each of the chemical reactions is performed according to the each of the derivative instruction sets, and optionally the initial instruction set, before the predictive model is updated. Preferably, each of the chemical reactions is performed according to the each of the derivative instruction sets, and optionally the initial instruction set, are carried out in parallel. That is, each of the chemical reactions is performed according to the each of the derivative instruction sets, and optionally the initial instruction set, are performed at the same time.

[0093] The initial instruction set may be obtained from one or more literature synthetic procedures, such as wherein the initial instruction is a literature synthetic procedure. The initial instruction set may be substantially the same as the literature synthetic procedure. The initial instruction set may be identical to the literature synthetic procedure. That is, the content of the instruction set is identical to the literature procedure. In some embodiments, the initial instruction set is obtained from literature synthetic procedures in natural language, reaction data and / or product data obtained from the analytical unit of the chemical synthesiser, or a combination thereof.

[0094] In some embodiments the initial instruction is a literature synthetic procedure.

[0095] 008889958The initial instruction set may be obtained by modifying one or more literature synthetic procedures. The initial instruction set may be obtained by generating an instruction set based on one or more related literature synthetic procedures. Optionally, the one or more literature synthetic procedures are for a different chemical reaction to the initial instruction set. The initial instruction set may be generated using a predictive model based on one or more different literature synthetic procedures.

[0096] As such, the method may also make extensive use of natural language chemical literature which may be directly executable on a chemical synthesiser. A chemist with no knowledge of programming can make use of synthetic operations to perform synthetic techniques based only on their knowledge of the synthetic procedures in the form of natural language.

[0097] Typically, a chemist need only provide the input data required for a target reaction product.

[0098] In some embodiments, the method is performed autonomously.

[0099] In the method, generating an optimised instruction set is made by providing the recorded reaction data and / or product data to the predictive model. The predictive model adjusts one or more reaction parameters to optimise a characteristic of the chemical reaction.

[0100] In some embodiments, the reaction data is selected from reaction colour, reaction temperature, reaction vessel pressure, reaction pH, reaction filter rate, surrounding humidity, surrounding pressure, surrounding temperature, or a combination thereof, preferably reaction colour, reaction temperature or a combination thereof.

[0101] The reaction data is recorded throughout the chemical reaction. That is, the reaction data is recorded from the start of the reaction to the end of the reaction. The start of the reaction may be taken as the time when the reactants are added to a reaction vessel. The end of the reaction may be taken as the time when the product is removed from the reaction vessel.

[0102] The reaction data may be recorded at regular intervals or continuously throughout the reaction. Preferably, the reaction data is recorded at a sample rate of 1 s-1or more, such as 10 s'1or more, such 100 s-1or more, such as 1000 s-1or more. By recording the reaction data at a higher sample rate the information density available when updating the predictive model is greater.

[0103] In some embodiments, the product data is spectroscopic data, such as UV-Vis, Infra-red, Raman, NMR, or mass spectrometry data; chromatography data, such as gas or high-performance liquid chromatography data; or a combination thereof.

[0104] In some embodiments, the reaction parameter for the instruction set is selected from reagent volume, reagent mass, reagent addition rate, solvent volume, solvent mass, solvent addition

[0105] 008889958rate, reaction temperature, rate of change of reaction temperature, reaction pressure, rate of change of reaction pressure, pH, stirring rate, reaction mixture colour and reaction time.

[0106] In the final step, an optimised instruction set is generated based on the updated predictive model. In some embodiments, the method may comprise performing a chemical reaction according to the optimised instruction set.

[0107] The present invention also provides a chemical synthesiser for carrying out the method of the first aspect. The chemical synthesiser comprises a plurality of reaction vessels to allow for a plurality of different reactions to be carried out in parallel. That is, the chemical reactions are performed at the same time.

[0108] In some embodiments, at least two of the chemical reactions performed according to the number of derivative instruction sets, and optionally the initial instruction set, are performed in parallel. Preferably, all the chemical reactions performed according to the number of derivative instruction sets, and optionally the initial instruction set, are performed in parallel.

[0109] In general, there is provided a chemical synthesiser for optimising an instruction set for a chemical reaction, the chemical synthesiser comprising a plurality of reaction vessels, an analytical unit and a control system, wherein:

[0110] the control system is for generating a number of derivative instruction sets from an initial instruction set using a predictive model;

[0111] the plurality of reaction vessels is for performing chemical reactions according to the derivative instruction sets and optionally the initial instruction set; and

[0112] the analytical unit is for recording reaction data during each of the chemical reactions and / or recording product data on the reaction product of each of the chemical reactions; and wherein the control system is for updating the predictive model using the reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction.

[0113] In a second aspect of the invention there is provided a chemical synthesiser for optimising an instruction set for a chemical reaction, the chemical synthesiser comprising a plurality of reaction vessels, an analytical unit and a control system, wherein:

[0114] the control system is for obtaining an initial instruction set for a chemical reaction, and for generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different;

[0115] the plurality of reaction vessels is for performing chemical reactions according to the number of derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performable in a different reaction vessel to produce a reaction product for each of the chemical reactions; and

[0116] the analytical unit is for recording reaction data during each of the chemical reactions and / or recording product data on the reaction product of each of the chemical reactions;

[0117] 008889958wherein the control system is for updating the predictive model using the reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction, and for generating an optimised instruction set based on the predictive model.

[0118] In some embodiments, the chemical synthesiser is for autonomously optimising the instruction set. The chemical synthesiser may be an autonomous chemical synthesiser.

[0119] In some embodiments, the control system comprises a control unit, and the predictive model is located within a control unit of the chemical synthesiser. Thus, the control unit may be suitably programmed with a predictive model to generate a number of derivative instruction sets from the initial instruction set.

[0120] In some embodiments, the control system comprises a control unit and a server in communication with the control unit, and the predictive model is stored on the server.

[0121] In some embodiments, the server is for obtaining an initial instruction set for a chemical reaction and for generating a number of derivative instruction sets from the initial instruction set using a predictive model, and the server sends the derivative instruction sets and optionally the initial instruction set to the control unit.

[0122] In some embodiments, the server is for receiving reaction data and / or product data of each of the chemical reactions from the control unit, and the server is for updating the predictive model using the reaction data and / or product data to optimise one or more characteristics of the chemical reaction, and for generating an optimised instruction set based on the predictive model, and optionally sending the optimised instruction set to the control unit.

[0123] The chemical synthesiser may be a robotic chemical synthesiser for autonomous performance of the chemical synthesis, and for autonomous control of the analytical device.

[0124] The control unit may be suitably programmed with an algorithm to analyse reaction data, product data and reaction parameters from previous test synthesis, and to adjust the reaction parameters for the next test synthesis.

[0125] The analytical unit may comprise a colour sensor, a temperature sensor, a pressure sensor, a pH sensor, a flow rate sensor, a humidity sensor or a combination thereof.

[0126] The analytical unit may comprise an infra-red spectrometer, a Raman spectrometer, a NMR spectrometer, a mass spectrometer, a gas chromatography instrument, a high-performance liquid chromatography instrument, or a combination thereof.

[0127] 008889958In some embodiments, the chemical synthesiser further comprises a bottom module and a top module, wherein:

[0128] the bottom module is for receiving the outside of the reaction vessel and is for applying reaction conditions of the instruction set to the outside of the reaction vessel, and the top module is for accessing the inside of the reaction vessel and for applying reaction conditions of the instruction set to the inside of the reaction vessel.

[0129] In some embodiments, the bottom module is for receiving multiple reaction vessels. In some embodiments, the bottom module is for receiving two or more reaction vessels, such as four or more reaction vessels, such as six or more reaction vessels. In this way, the bottom module is for applying the same conditions to multiple reaction vessels. The bottom module may be for applying the same conditions to a set of two or more, such as four or more, such as six or more reaction vessels.

[0130] In some embodiments, the top module is for accessing the inside of one reaction vessel. In this way, the top module is for applying different conditions to each vessel.

[0131] The combination of the top and bottom modules allow for each reaction vessel to be reacted under different conditions.

[0132] The derivative instruction sets may be generated to be compatible with the chemical synthesiser. The step of generating the derivative instruction sets may take into consideration the configuration of the chemical synthesiser.

[0133] For example, where the chemical synthesiser has a bottom module for receiving multiple reaction vessels, and is for applying the same conditions to multiple reaction vessels, the derivative instructions sets may be generated such that conditions which are applied by the bottom module are the same for multiple reaction vessels.

[0134] In the same way, where the chemical synthesiser has a top module for receiving each reaction vessels, and is for applying different conditions to each reaction vessels, the derivative instructions sets may be generated such that conditions which are applied by the top module are the different for each reaction vessels which have the same conditions applied by the bottom module. In this way, the top module and the bottom module may provide different conditions to each reaction vessel.

[0135] The chemical synthesiser may also be a part of a chemical synthesis platform. The chemical synthesis platform is typically a network of multiple chemical synthesisers.

[0136] In a third aspect of the invention there is provided a chemical synthesis platform comprising the chemical synthesiser of the second aspect and one or more additional chemical

[0137] 008889958synthesisers, wherein at least one of the additional chemical synthesisers are different to the chemical synthesiser.

[0138] In some embodiments, at least one of the additional chemical synthesisers has one reaction vessel. That is, the additional chemical synthesisers may have only one reaction vessel.

[0139] In a fourth aspect of the invention there is provided a use of the chemical synthesiser of the second aspect or the chemical synthesis platform of the third aspect, for optimising an instruction set for a chemical reaction.

[0140] In some embodiments, the use is for reducing the time to optimise an instruction set for a chemical reaction. The time may be reduced compared to a sequential optimisation procedure.

[0141] In some embodiments, the use is for reducing the resources for optimising an instructions et for a chemical reaction. The resources may include reactants or solvents. The resources may include lab space. The resources may include user time, such as user time interacting with the chemical synthesiser or the frequency of user interaction with the chemical synthesiser. The resources may be reduced compared to a sequential optimisation procedure.

[0142] These and other aspects and the embodiments of the invention are described in further detail herein.

[0143] Summary of the Figures

[0144] The present invention is described with reference to the figures listed below.

[0145] Figure 1 shows an overview of an example optimisation procedure of the invention, for optimising a chemical reaction. The procedure involves a condition prediction and a reaction estimator to determine the complexity of a chemical reaction. The Design of Experiments step determines the number of derivative instruction sets to be generated based on the complexity.

[0146] Figure 2 shows an example visual representation of the execution of a number of derivative instruction sets generated from an initial instruction set. Each node is a unique chemical reaction, and the colour captures the complexity of each step of the chemical reaction, where the complexity is based on a prediction of the probability that at least one derivative instruction set will produce a reaction product with a desired characteristic.

[0147] 008889958Figure 3 shows an example overview of the relationship between the complexity of a chemical reaction and the number of derivative instruction sets to be generated and, by extension, the number of chemical reactions to be performed.

[0148] Figure 4 shows an overview of an example chemical synthesiser used for the methods of optimisation.

[0149] Figure 5 shows a regression plot (top) and a classification plot (bottom) used to correlate the complexity values of 545 different types of chemical reactions, where the complexity is manually assigned by chemists (‘true label) and predicted by an XGBoost machine learning model (‘predicted label’)..

[0150] Figure 6 shows a CAD rendering (top) of an example chemical synthesiser and a photograph of an example chemical synthesiser (bottom).

[0151] Figure 7 shows a CAD rendering (top) of an example bottom module of a chemical synthesiser and a photograph of an example bottom module (bottom) of a chemical synthesiser.

[0152] Detailed Description of the Invention

[0153] Generally, the present invention relates to a method of optimising an instruction set for a chemical synthesis. The method may be performed on a chemical synthesiser. The method is preferably performed autonomously.

[0154] Optimisation refers to a change or a series of changes which results in an improvement. That is, a change which results in the achievement of a desired result.

[0155] In the context of a chemical synthesis, an optimisation is a change to the synthesis method which may improve a quality of the chemical synthesis itself or a quality of the reaction product. Qualities of the chemical synthesis and qualities of the reaction product may be known as characteristics of the synthesis. A user may determine which characteristics of the synthesis are to be improved.

[0156] For the reaction product, optimisation of a chemical synthesis may result in an increase in reaction product yield, reaction product purity, reaction product to starting material ratio, or reaction product to impurity ratio. For the reaction itself, optimisation may result in a reduction in solvent mass consumption, reagent mass consumption. It could also reduce the use of environmentally harmful reagents or solvents, or expensive reagents or solvents.

[0157] A chemical synthesis may be defined by an instruction set. The instruction set may be a versionable executable code which is capable of execution by a chemical synthesiser,

[0158] 008889958having all the abstract explicit operations in a chemical programming language. The instruction set is an apparatus agnostic description of chemical operations, and the system interprets the instruction set to execute a synthesis on a chemical synthesiser.

[0159] Accordingly, in optimising a chemical synthesis the method of the present invention is a method of optimising an instruction set for a chemical synthesis.

[0160] In a general aspect, the present invention provides a method of optimising a chemical synthesis, the optimisation comprising the steps of:

[0161] (i) obtaining an initial instruction set for a chemical reaction,

[0162] (ii) generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different,

[0163] (iii) performing chemical reactions according to the initial instruction set and / or each of the derivative instruction sets, wherein each of the chemical reactions is performed in a chemical synthesiser to produce a reaction product for each of the chemical reactions; and (iv) recording reaction data using an analytical unit of the chemical synthesiser during each of the chemical reactions and / or recording product data on the reaction product of each of the chemical reactions;

[0164] (v) updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction; and

[0165] (vi) generating an optimised instruction set based on the predictive model.

[0166] The work in the present case also exemplifies the optimisation of a chemical synthesis where the derivative instruction set for the chemical synthesis may be automatically generated from literature procedures using an intelligent software system based on the open-source universal chemical programming language standard and format, xDL. This open standard has been designed to allow any chemical transformation to be precisely expressed and run on any compatible robotic platform. Thus, described herein is a method and chemical synthesiser which can generate all the components necessary to execute and optimise an automated synthesis directly from a literature procedure using a universal chemical description language.

[0167] Some synthesis methods and synthesis apparatus have been described previously.

[0168] WO 2024 / 091573 describes systems and methods for generating reaction conditions using machine learning and robotic experimentation. The method involves selecting conditions for a reaction of molecules based on historic use and the structural and functional diversity of the molecules. The method then performs reactions based on these conditions, and optimises the reaction conditions using a machine learning model. The document does not describe an optimisation approach involving generating a number of derivative instruction

[0169] 008889958sets and, by extension, the number of reactions to be performed, based on the complexity of the reaction.

[0170] US 2021 / 125060 describes optimising experimental conditions using a neural network. In the optimisation, structural information of reactants and products are provided to a prediction model to generate combinations of experimental conditions for generating the product. A predicted yield is calculated for each condition, so that an experiment priority can be determined based on the most optimal predicted yield. An experiment is performed based on that priority, and the prediction model is updated based on feedback from that experiment. The document determines one experimental priority based on the predicted yield, and performs the experiment based on that single experimental priority to update the experimental condition. The document does not describe performing multiple reactions in line with a number of derivative instruction sets generated from an initial instruction set.

[0171] WO 2023 / 131726 describes an automated method for exploring chemical space and looking for chemical products having desirable physical or chemical properties, and an apparatus for performing that method. The method involves preparing a library of products using different reaction conditions, and selecting products from that library based on desirable physical or chemical property. The selected product is then used as a starting material in the synthesis of a second library of products, using different reaction conditions. From that second library, a further desirable product may be selected. The document does not describe the generation of derivative instruction sets from an initial instruction set, nor that the number of derivative instruction sets corresponds to the number of chemical reactions performed. The document also does not describe that the number of products to be synthesised is based on the complexity of the reaction.

[0172] US 5463564 describes a computer-based, iterative process for generating chemical products having defined properties. The process involves generating a chemical library and performing reactions to generate chemical products, analysing those chemical products to identify compounds with desired properties, and generating new synthesis instructions for a next iteration. The document does not describe that synthesis instructions are derived from an initial instruction set, nor that the number of products to be synthesised in the library is based on the complexity of the initial instruction set.

[0173] US 2020 / 0225251 describes an apparatus and method for reaction screening and optimisation. The method involves providing experimental design parameters to a system for controlling reactions within reaction vessels, performing reactions within the reaction vessels, and analysing compositions and identifying an optimum reaction condition. The document does not describe that synthesis instructions are derived from an initial instruction set, nor that the number of products to be synthesised in the library is based on the complexity of the initial instruction set.

[0174] 008889958Du et al., Salley et al. and He et al. each generally describe automated robotic synthesis platforms which use algorithms to optimise reactions. None of these documents describe generating a number of derivative instruction sets from an initial instruction set. None of these documents describe that the number of reactions performed is based on the complexity of the reaction.

[0175] Method of Optimising a Chemical Reaction

[0176] In a first aspect of the invention there is provided a method of optimising an instruction set for a chemical reaction, the method comprising the steps of:

[0177] (i) obtaining an initial instruction set for a chemical reaction;

[0178] (ii) generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different;

[0179] (iii) performing chemical reactions according to the number of derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performed in a different reaction vessel of a chemical synthesiser;

[0180] (iv) recording reaction data during each of the chemical reactions and / or recording product data on a reaction product of each of the chemical reactions using an analytical unit of the chemical synthesiser;

[0181] (v) updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction; and

[0182] (vi) generating an optimised instruction set based on the updated predictive model.

[0183] The methods of the invention are for optimising an instruction set for a chemical reaction. That is, the instruction set is optimised to improve a characteristic of the chemical reaction.

[0184] The chemical reaction is performed in a chemical synthesiser, such as an autonomous chemical synthesiser (e.g., a chemical robot). The chemical synthesis is typically performed on a chemical synthesiser, as described herein. The chemical synthesis may be mapped into a reactionware of a chemical synthesiser.

[0185] In some embodiments, the method is performed autonomously. The method may be performed by a chemical synthesiser operating autonomously, where the chemical synthesiser is described herein.

[0186] The chemical synthesis is not particularly limited and may encompass methods for the production of small organic compounds, metal complexes, supramolecular structures and polymers, amongst others. The chemical synthesis may also encompass biological reactions, or reactions where biomolecules participate as reagents or catalysts. Such biomolecules may include polypeptides, such as proteins, including enzymes and antibodies, polynucleotides, polysaccharides, metabolites and cofactors, amongst many others.

[0187] 008889958The chemical synthesis may also include one or more steps of preparing a reaction mixture for a chemical reaction, as well as one or more methods for the work-up of a reaction mixture following a chemical reaction, and one or more methods for purifying a product of a chemical reaction. The optimisation may focus on the chemical reaction itself, as well as the steps of preparing the reaction mixture, work-up and purification. For each chemical reaction in a chemical synthesis, the chemical reaction may include one or more work-up, purification or preparation steps.

[0188] The chemical synthesis may be a single step synthesis. Alternatively, the chemical synthesis may be a multistep synthesis. The multistep synthesis may comprise two or more steps, such as a step of preparing an intermediate compound from one or more reagents, and the subsequent transformation of that intermediate compound into a reaction product. The multistep synthesis may comprise parallel steps, such as a step of preparing two intermediate compound in parallel, and a step of reacting the intermediate compounds. The chemical synthesis may comprise two or more chemical reactions, such as three or more chemical reactions. The chemical synthesis may comprise a series of chemical reactions, as well as well as parallel chemical reactions, that are later brought to convergence for the production of a target reaction product.

[0189] The chemical synthesis may include a preparation step, the preparation step comprising preparing a chemical synthesiser for performance of a reaction, including the steps of generating an instruction set, generating a virtual reaction platform for execution of the instruction set, and generating the physical hardware (such as a physical reactionware) for the reaction according to the instruction set from the virtual platform.

[0190] By way of illustration, the worked examples provided in the present case exemplify the method of optimising a Van Leusen oxazole chemical synthesis in accordance with the parallel optimisation methods, and this is compared with a similar chemical synthesis in accordance with an iterative optimisation approach. As described in the worked examples, the parallel optimisation method is more efficient in term of time and resources, while also achieving a reaction yield which is the same as or better than the iterative optimisation approach.

[0191] Instruction Sets

[0192] An instruction set comprises a list of the individual coded operations needed to perform a chemical synthesis. The instruction set may be a versionable executable code which is capable of execution by a chemical synthesiser, having all the abstract explicit operations in a chemical programming language. The instruction set is an apparatus agnostic description of chemical operations, and the system interprets the instruction set to execute a synthesis on a chemical synthesiser.

[0193] 008889958The instruction set may be in natural language, or in any suitable machine-readable format. Typically, the instruction set is in a mark-up language, such as a descriptive mark-up language, such as a chemical descriptive mark-up language, for example XML or XDL. The preferred format for the instruction set is a mark-up language adapted for chemical synthesis, such as XDL. XDL is described in (Steiner). XDL is a universal chemical programming language standard and format which allows end-to-end chemical procedures to be described in a general, platform-independent manner.

[0194] The instruction set is an instruction set for the chemical synthesis that can be adjusted to provide an optimised instruction set. After performing a number of chemical reactions, the instructions set is optimised to provide an optimised instruction set. The optimised instruction set is typically the instruction set where the desired characteristic of the reaction product (e.g., yield, purity, conversion) is improved.

[0195] The instruction set provides is a description of a method, or synthetic procedure, of making a chemical product. A synthetic procedure typically comprises the different experimental operations or steps required to produce the product. It may comprise information on the different reagents (e.g. starting materials and reactants), solvents and catalysts, and the quantities required, to produce the product. It may comprise information on any specialist equipment needed. However, it is typical for a synthetic procedure to omit certain specific information that may be deemed obvious to a skilled chemist. Typically, information on the necessary equipment is omitted. Similarly, information on work-up (e.g. quenching of the reaction mixture), flushing (e.g. with inert gases), washing, extraction and purification steps is also omitted.

[0196] The instruction set may be provided via manual input into a predictive model by a human user, such as a human user providing an instruction set electronically to the predictive model.

[0197] The instruction set may be provided by extracting the synthetic procedure from literature.

[0198] A synthetic procedure for an instruction set is typically found in an academic journal article in the chemical, biological or materials sciences. It may be located either in the main text of the article, or in the associated supplementary information. A synthetic procedure may also be found in other academic texts, such as a review article, letter or book. It may be found in a specialised synthetic textbook or a laboratory standard operating procedure. Dedicated web repositories for synthetic procedures are also known (e.g. http: / / www.orgsyn.org). They may also be found in teaching materials, such as a general textbook or a laboratory procedure. The procedure may also be set out in a thesis, or a company report.

[0199] 008889958Synthetic procedures are typically in natural language. That is, in free, unstructured text. For example, in prose, such as in encrypted prose. These natural language synthetic procedures are suitable for use in the present invention.

[0200] The method may comprise an optional pre-processing step. The pre-processing stage can be used to simplify and shorten the natural language synthetic procedure to make the subsequent processing stages quicker. Several pre-proceedings steps may be included. The pre-processing steps are performed automatically.

[0201] The pre-processing step may comprise performing optical character recognition (OCR) on the document displaying the natural language synthetic procedure. OCR converts an image of text, such as scanned text or a photograph of text, into machine-encoded text. Thus, including OCR enables the method to use scanned or photographed synthetic procedures from physical (hard-copy) articles or handwritten notebooks.

[0202] The pre-processing step may comprise removing HTML from the synthetic procedure.

[0203] HTML code is typically included in synthetic procedures contained in an electronic database. For example, synthetic procedures in the Reaxys database typically contains HTML code. It is not necessary to understand the HTML in order to extract an instruction set from the synthetic procedure.

[0204] The pre-processing step may comprise removing formatting markings from the synthetic procedure. For example, the pre-processing steps may comprise removing formatting marks indicating italicisation, emboldening and underling.

[0205] The pre-processing step may comprise removing non-printing characters from the synthetic procedure. For example, the pre-processing step may comprise removing tabs and paragraph markers (carriage returns). Non-breaking characters may be replaced with standard counterparts (e.g. replacement of a non-breaking space with a space).

[0206] The pre-processing step may comprise removing analytical data from the synthetic procedure. Analytical data includes, for example, NMR, IR, melting point and mass spectrometry data. This analytical data can be useful for verifying the identity of chemical compounds, for example, the final product. This analytical data may be termed characterisation data. Characterisation data is typically found at the end of a synthetic procedure, or at the end of each step within a synthetic procedure. However, characterisation data is not necessarily required for the extraction of the instruction set. Removing the characterisation data in a pre-processing step reduces the text, speeding up subsequent steps.

[0207] Certain analytical data may be useful for performing the synthetic procedure. For instance, analytical data defining the end-point of a reaction or synthetic operation. For example,

[0208] 008889958where a synthetic procedure requires a synthetic operation to be performed until a given colour change is observed, or until a certain peak within the chromatograph of the reaction mixture disappears. This analytical data may also include the appearance or disappearance of specific peaks within an NMR or IR spectrum of the reaction mixture. This analytical data may be termed reaction-monitoring data. Reaction monitoring data is typically found embedded within the text of the natural language synthetic procedure. Optionally, this analytical data may be retained in the synthetic procedure.

[0209] The pre-processing may comprise normalising the text. This transforms the text into a consistent format. For example, normalising the text may comprise inserting missing spaces after full stops or before certain units, such as “°C”.

[0210] The pre-processing step may comprise translating the text, for example, translating the text from German, French, Chinese, Japanese, Korean or Russian into English. The method may use a single language, preferably English, for standardisation within the system.

[0211] The pre-processing step may include de-encrypting the synthetic procedure. Any suitable de-encryption may be used.

[0212] Finally, the pre-processing step may include making specific replacements to standardize the appearance of certain phrases. For example, words denoting numbers may be replaced with their respective numerals (e.g. replacement of “ninety nine” with “99”). Full stops that denote abbreviations may be removed (e.g. replacement of “min.” in the phrase “5 min.” with “min” to give “5 min”). The different characters used to represent the degree symbol may be replaced with the standard symbol (e.g. replacement of the letter o in superscript “°” or “°” or the ordinal indicator “°” with the degree symbol “°”).

[0213] The method comprises providing an initial instruction set for a chemical reaction. The initial instruction set corresponds to a set of instructions for a chemical synthesiser to perform, where the intention may be to optimise the characteristics of the chemical reaction defined by the initial instruction set. The initial instructions set typically comprises reaction parameters, such that these reaction parameters may be adjusted and optimised to obtain a desired characteristic of the chemical reaction.

[0214] Typically, the reaction product of interest as defined by the initial instruction set for the chemical reaction is a reaction product in which the desired characteristics are intended to be optimised. The characteristics may be those as described further herein.

[0215] The method of the invention seeks to provide an approach towards the optimisation of chemical synthesis which is more time-efficient and resource-efficient. The method therefore involves generating a specific number of derivative instruction sets using a predictive model.

[0216] 008889958Typically, the predictive model analyses the initial instruction set, for example based on its complexity as described herein, and generates a number of derivative instruction sets which define the chemical space around the chemical reaction defined by the initial instruction set. Any predictive model may be used, which is capable of generating a number of derivative instruction sets by adjusting reaction parameters.

[0217] The generation is made based on an increase or decrease to one of the reaction parameters. For example, an increase or decrease to the temperature of the chemical reaction, an increase or decrease to the duration of the chemical reaction, or an increase or decrease to the type of starting material.

[0218] The generation is made based on a prediction that the adjustment to generate at least one derivative instruction set will improve the characteristics of the chemical reaction.

[0219] The predictive model is typically a machine learning predictive model. Any suitable machine learning predictive model may be used, which is capable of generating a number of derivative instruction sets by adjusting reaction parameters. The predictive model may use manual search, grid search, random search, genetic model or Bayesian model-based optimization (e.g., SMBO). The predictive model may use gradient-boosted decision treebased optimisation (e.g., GBDT). Preferably, the predictive model is a random search, genetic model, GBDT, or SMBO algorithm. Preferably, the predictive model is a GBDT algorithm.

[0220] Where the predictive model is a random search algorithm, the values of the reaction parameters are determined from a random number array (sampled from a uniform distribution) within the parameters constraints.

[0221] Where the predictive model is a genetic model algorithm, it may use truncation selection, single point crossover, and random reset mutation while preserving the best solution (elitism). In case of premature convergence, the population is typically reinitialized. The genetic algorithm may include hyperparameters selected from population size (e.g., number of individuals in the population) and probability of change (e.g., probability of mutating a gene).

[0222] Where the predictive model is GBDT, the algorithm may be implemented using a XGBoost machine learning model, as described in the examples.

[0223] The derivative instruction sets may be generated to be compatible with the chemical synthesiser, in particular the reaction vessels of the chemical synthesiser. The step of generating the derivative instruction sets may take into consideration the configuration of the chemical synthesiser.

[0224] 008889958For example, where the chemical synthesiser has a bottom module for receiving multiple reaction vessels, and is for applying the same conditions to multiple reaction vessels, the derivative instructions sets may be generated such that conditions which are applied by the bottom module are the same for multiple reaction vessels.

[0225] Where the chemical synthesiser has a top module for receiving each reaction vessels, and is for applying different conditions to each reaction vessels, the derivative instructions sets may be generated such that conditions which are applied by the top module are the different for each reaction vessels which have the same conditions applied by the bottom module. In this way, the top module and the bottom module may provide different conditions to each reaction vessel.

[0226] Complexity

[0227] Preferably, the number of derivative instruction sets is determined based on the complexity of the chemical reaction defined by the initial instruction set. Therefore, the method may comprise determining the complexity of the initial instruction set. The method may then comprise calculating the number of derivative instruction sets to be generated based on the complexity.

[0228] A predictive model is used to generate derivative instruction sets, and to determine the number of derivative instruction sets to be generated. Each of the derivative instruction sets generated by the predictive model may be different. Each of the derivative instruction sets generated by the predictive model may also be different to the initial instruction set.

[0229] Preferably, at least one derivative instruction set, where the chemical reaction according to the derivative instruction set is performed, can produce at least one chemical reaction, such as at least one reaction product, with a desired characteristic.

[0230] Typically, the greater the complexity of a chemical reaction, the greater the number of derivative instruction sets to be generated.

[0231] The complexity of the initial instruction set may be based on a prediction of the probability that performing a chemical reaction according to the initial instruction set will result in a chemical reaction with a desired characteristic being performed.

[0232] The complexity of the initial instruction set may be based on a prediction of the probability that performing a chemical reaction according to the initial instruction set will produce a reaction product with a desired characteristic.

[0233] The complexity of a chemical reaction may be defined as the probability that performing a chemical reaction will result in a chemical reaction with a desired characteristic being

[0234] 008889958performed. The desired characteristic may be a characteristic of the reaction parameter of the chemical reaction, or a characteristic of the reaction product of the chemical reaction, or a combination thereof. Typically, the desired characteristic is a characteristic of the reaction product.

[0235] A more complex chemical reaction typically results in a lower probability of the chemical reaction producing a reaction product with a desired characteristic. Contributing factors which may determine the complexity of a chemical reaction may include the structural difference between the starting material and the reaction product, the type of chemical reaction that is required to be performed to obtain the reaction product, and the sensitivity of the chemical reaction to changes in the reaction parameters.

[0236] In some embodiments, the method further comprises determining the complexity of the initial instruction set using the predictive model, and calculating the number of derivative instruction sets to be generated based on the complexity.

[0237] The complexity of the initial instruction set is typically dependent on the complexity of the chemical synthesis. Typically, the complexity of the chemical synthesis and, by extension, the complexity of the initial instruction set may directly correlate with the number of derivative instruction sets to be generated.

[0238] In some embodiments, the complexity of the initial instruction set is based on:

[0239] (a) the degree of structural difference between the starting material and the reaction product of the chemical reaction;

[0240] (b) the reaction type of the chemical reaction;

[0241] (c) the reaction mechanism of the chemical reaction; and / or

[0242] (d) literature reports of the chemical reaction.

[0243] The factors of which the complexity of the initial instruction set may be based are described as follows.

[0244] The degree of structural difference between the starting material and the reaction product of the chemical reaction may be quantified by the number of functional groups in the starting material, the number of functional groups in the reaction product, or the difference in the number of functional groups between the starting material and the reaction product.

[0245] Typically, a greater number of functional groups in either material may be expected to increase the complexity relative to a lower number of functional groups in either material. For example, certain functional groups may need to be selective and / or reactive in the reaction.

[0246] 008889958Typically, a greater difference in the number of functional groups between the starting material and the reaction product may also be expected to increase the complexity relative to a lower difference. The complexity may increase, relative to the same number of functional groups in both materials, if there is a greater number of functional groups in the reaction product relative to the starting material, for example because a number of functional groups are introduced during the reaction. The complexity may also increase, relative to the same number of functional groups in both materials, if there is a greater number of functional groups in the starting material relative to the reaction product, for example because a number of functional groups are removed during the reaction.

[0247] The degree of structural difference between the starting material and the reaction product of the chemical reaction may be quantified by the number of starting materials, the number of reaction products, or the difference between the number of starting materials and the number of reaction products.

[0248] The reaction type of the chemical reaction may provide an indication for the complexity of the initial instruction set. Where the reaction type matches that of a known, well-established reaction type having a known, predetermined complexity, the complexity of the initial instruction set may be lower compared to where the reaction type does not match a known reaction type. Examples of known, well-established reaction types may include, but are not limited to, Suzuki coupling, amide coupling, reductive amination, carboxylic acid esterification, ester hydrolysis, Sonogashira coupling, and double bond reduction.

[0249] The reaction type may be dependent on the reaction parameters, as described herein. For example, the reaction type may be dependent on the reagent, catalyst, solvent, or any other component, or may be dependent on the temperature or the pressure. Where a reagent, catalyst, solvent, or any other component require specialised handling, the complexity of the reaction may be greater compared to where specialised handling is not required. Where the reaction requires specialised atmosphere, such as an inert atmosphere, the complexity of the reaction may be greater compared to performing under standard atmospheric conditions.

[0250] The reaction mechanism of the chemical reaction may provide an indication for the complexity of the initial instruction set. Factors of the reaction mechanism reaction may include the stability of transition states during the reaction, the steric hindrance around reaction functional groups of the starting material, and potential competing reactions.

[0251] The literature reports of the chemical reaction may provide an indication for the complexity of the initial instruction set. The initial instruction set, which defines a chemical reaction, may be compared against a literature report of a chemical reaction to identify a similarity between the initial instruction set and chemical reactions in the literature. The literature reports may be obtained from chemical databases which may be known to the user.

[0252] 008889958A literature report may include information about an existing reaction. Thus, for example, yield provided in literature reports may be used to determine complexity. A high average yield reported may indicate that a reaction is less complex than a low average yield. In addition, or alternatively, a high variance in the average yield reported may indicate that a reaction is more complex than a low variance in the average yield.

[0253] The complexity of the initial instruction set may be based on any combination of the above factors. The factors may be weighted depending on their importance in the complexity prediction. Where the complexity of the initial instruction set is high, a larger variance and spread of reaction parameters may be required for exploration of the chemical space to identify an optimised instruction set. The factors of which the complexity of the initial instruction set may be based may be directly correlated with the number of derivative instruction sets to be generated. Put simply, the more complex the initial instruction set, the more derivative instructions sets to be generated, and thus the more chemical reactions to be performed.

[0254] In some embodiments, each of the derivative instruction sets generated by the predictive model is different. For example, each of the derivative instruction sets comprises a different reaction parameter, such as a different chemical condition, a different physical condition, a different chemical input, or a different physical input.

[0255] In some embodiments, the complexity of the initial instruction set is based on a prediction that the initial instruction set will provide a product with a desired characteristic.

[0256] In some embodiments, performing at least one chemical reaction according to at least one derivative instruction set produces at least one chemical reaction with a desired characteristic.

[0257] In some embodiments, performing at least one chemical reaction according to at least one derivative instruction set produces at least one reaction product with a desired characteristic.

[0258] The complexity of the initial instruction set may be the complexity of the initial instruction set for each chemical reaction in the synthetic pathway, where the synthetic pathway comprises a series of chemical reactions. In other words, a synthetic pathway may correspond to a multi-step synthesis defined by a series of chemical reactions. Each chemical reaction in the synthetic pathway may be defined by an initial instruction set. Each complexity of each initial instruction set of each chemical reaction in the synthetic pathway may be the same or different. The complexity of each initial instruction set may be based on any combination of the above factors. The factors may be weighted depending on their importance in the complexity prediction and in the initial instruction set.

[0259] 008889958Availability of Chemical Synthesiser

[0260] Preferably, the number of derivative instruction sets is determined based on the availability of the chemical synthesiser for performing the optimisation method. Therefore, the method may comprise determining the availability of the chemical synthesiser. The method may then comprise calculating the number of derivative instruction sets to be generated based on the availability.

[0261] As described above, a predictive model is used to generated derivative instruction sets.

[0262] Typically, the greater the availability of the chemical synthesiser, the greater the number of derivative instruction sets to be generated.

[0263] The availability of the chemical synthesiser may be defined as the extent to which the chemical synthesiser, or its constituent units thereof, are capable of being used to perform chemical reactions at a given time. The availability of the chemical synthesiser may therefore define an upper limit on the number of chemical reactions that may be performed concurrently or within a given optimisation campaign.

[0264] A chemical synthesiser of reduced availability typically results in a lower number of derivative instruction sets to be generated. A chemical synthesiser of greater availability typically results in a larger number of derivative instruction sets to be generated.

[0265] In some embodiments, the method further comprises determining the availability of the chemical synthesiser.

[0266] This ensures that the chemical reactions to be performed based on the derivation instruction sets are guaranteed to be executable and performable on the available chemical synthesiser, without requiring operator intervention. This step is particularly beneficial for parallel reaction optimisations, where the chemical synthesiser may be required to perform multiple chemical reactions simultaneously, and autonomously.

[0267] Where a number of derivative instruction sets is initially determined by the complexity of the initial instruction sets, that number of derivative instruction sets may be cross-checked with the availability of the chemical synthesiser for performing the optimisation method.

[0268] In some embodiments, the availability of the chemical synthesiser is based on:

[0269] (a) the number of available reaction vessels of the chemical synthesiser;

[0270] (b) the amount of available starting material;

[0271] (c) the operational status of the chemical synthesiser; and / or

[0272] (d) limitations defined by an operator

[0273] 008889958The factors of which the availability of the chemical synthesiser may be based are described as follows:

[0274] In some embodiments, the availability of the chemical synthesiser is determined based on the number of available reaction vessels of the chemical synthesiser. Where there is a maximum number of reaction vessels available for performing chemical reactions, the number of derivation instruction sets may not exceed that number.

[0275] In some embodiments, the availability of the chemical synthesiser is determined based on the amount of available starting material. Where there is a maximum amount of available starting material, the number of derivation instruction sets may be selected such that the maximum number of chemical reactions to be performed may not require more starting material than is available.

[0276] In some embodiments, the availability of the chemical synthesiser is determined based on the operation status of the chemical synthesiser. For example, the chemical synthesiser may have, at the time of the optimisation method, reaction vessels which may be unavailable for use, due to cleaning or maintenance.

[0277] In some embodiments, the availability of the chemical synthesiser is determined based on the limitations defined by an operator. For example, an operator may explicitly limit the number of derivative instruction sets to be generated, and thus the number of chemical reactions to be performed.

[0278] Preferably, the number of derivative instruction sets to be generated is based on the complexity of the initial instruction set and the availability of the chemical synthesiser.

[0279] In some embodiments, the number of derivative instruction sets to be generated is determined by first determining the availability of the chemical synthesiser, then determining the complexity of the initial instruction set.

[0280] In this way, the availability of the chemical synthesiser may serve to provide an indication of a maximum number of derivative instruction sets to be generated, and thus a maximum number of chemical reactions to be performed.

[0281] In other embodiments, the number of derivative instruction sets to be generated is determined by first determining the complexity of the initial instruction set, then determining the availability of the chemical synthesiser.

[0282] In this way, the complexity of the initial instruction set may serve to provide an indication of a maximum number of derivative instruction sets to be generated, and thus a maximum number of chemical reactions to be performed.

[0283] 008889958Reaction Parameters

[0284] An instruction set comprises one or more, and typically a plurality of, reaction parameters. The reaction parameters together provide the information needed to perform a chemical synthesis.

[0285] Some reaction parameters may be fixed reaction parameters, in other words, reaction parameters which are not adjustable and so are not changeable between versions. Some reaction parameters may be variable reaction parameters, in other words, reaction parameters which are adjustable and so are changeable between versions, in order to optimise the chemical synthesis.

[0286] The instruction set typically comprises fixed reaction parameters and variable reaction parameters, preferably wherein the variable reaction parameters are variable between an upper limit and lower limit.

[0287] The upper limit and lower limit is each typically the range within which the reaction parameter can be adjusted to during the adjustment step. The upper limit and lower limit may be predetermined and input by an operator, or from the literature synthesis.

[0288] Alternatively, the variable range may be automatically determined based on the initial instruction set. For example, the upper limit and lower limit may be taken as + / - 30% of the reaction parameter provided in the initial instruction set. The upper limit and lower limit may also be limited by the boiling and freezing points of the fluid chemical inputs supplied to the reaction space, and the fluid product output.

[0289] In some embodiments, each of the reaction parameters in the instruction set is a variable reaction parameter. Preferably, all of the reaction parameters in the instructions set are variable reaction parameters.

[0290] Reaction parameters may also be referred to reaction conditions (such as chemical or physical conditions) or reaction inputs (such as chemical or physical inputs).

[0291] The reaction parameter may include physical parameters, such as heating, cooling, light, and mixing.

[0292] Physical parameters are intended to refer to a parameter that is not a material such as a reagent, catalyst, solvent, or a component. A physical parameter may refer to, for example, a temperature, such as the temperature of a particular chemical input, or the temperature of the reaction mixture, or the temperature of the reaction headspace, or the temperature of the surroundings. A modulation in temperature may refer to a physical parameter than can raise and / or lower temperature. A series of temperature parameters may be provided that is a

[0293] 008889958gradient of temperature increase and / or decreases. The range of temperature parameters may be limited by the boiling and freezing points of the fluid chemical inputs supplied to the reaction space, and the fluid product output. It is noted, however, that the reaction space may be suitably pressurised thereby to effectively alter the boiling and freezing points of the fluid chemical parameters. In this way a greater range of temperature parameters may be used.

[0294] Temperature parameter may be used to initiate reagents or favour certain reaction pathways.

[0295] Temperature parameter may also be used to investigate the stabilities of the reagents and products.

[0296] The physical parameter may be light. A series of light parameters may be provided that differ in one or more of intensity, wavelength, exposure time and spectrum. Light parameters may be used to initiate reagents or may be used to favour or alter certain reaction pathways. Light inputs may include UV-vis inputs. The physical parameter may be microwave radiation.

[0297] The physical parameter may be ultrasound. Such may be useful for the generation of reagents or products. Ultrasound may also aid the dissolution of material.

[0298] The physical parameter may be pressure. Pressure changes may be used to alter, for example, solvent boiling points.

[0299] The physical parameter to the system may be a process related parameter for the reaction mixture. The physical parameter may be a stirring rate of the reaction mixture. Thus, the parameter may be a time limited feature for reaction or admixture. After a set time, the reaction mixture may be analysed and the product quantified. Thus, reaction time may be a parameter. Similarly, other process features such as concentration and ratio of chemical parameters, such as the reagent and catalyst chemical inputs, may be a physical parameter.

[0300] The reaction parameter may include chemical parameters, such as the addition or removal of reagents, solvents and catalysts, or the rate of addition or removal of reagents, solvents and catalysts. Typically, the reference to a chemical parameter is a broad reference to the addition or removal of material, which may be a reagent, catalyst, solvent, or a component, that may allow the preparation of a reaction product.

[0301] The chemical parameter may specify if a chemical material is provided as a solid, or in a fluid for transfer to the reaction vessel.

[0302] 008889958Where the material is a fluid, it may be supplied in this form to the reaction space.

[0303] Alternatively, the material may be diluted, dissolved or suspended in a fluid for delivery to the reaction space. Thus, the material may be in solution or suspension. The fluid that dissolves or suspends the material is not particularly limited, and may be water or an organic solvent, for example. The fluid may be independently deliverable to the reaction space. The fluid is also used to provide separation between individual combinations of chemical inputs that are supplied to the reaction space thereby preventing contamination between different combinations.

[0304] The identity of the material will be dependent upon the reaction and formulation steps that are to be employed, and may be limited - though not necessarily - by the intended chemical synthesis.

[0305] A chemical material may be a reagent. A range of reagents may be provided that differ in their structure and functionality.

[0306] A chemical material may be a catalyst. A range of catalysts may be provided that differ in their activity, selectivity, or morphology.

[0307] A chemical material may be an acid or a base. A range of different acids and bases may be provided, where the acidity differs. Organic and inorganic acids and bases may be selected. Weak and strong acids and bases may be provided. A buffer may be used to maintain the reaction mixture at a pH range.

[0308] A chemical material may be a solvent. Organic solvents and water may be used. A range of non-polar, protic and aprotic solvents may be provided. In one embodiment, water is provided as a chemical input.

[0309] A chemical material may be a salt. A range of different salt forms of a particular component may be used. A range of organic and inorganic salts may be provided.

[0310] A chemical material may also be a gas. In some embodiments, a chemical material may be an inert gas, such as nitrogen or argon, to supply to the reaction space. In other embodiments, the chemical input is a reaction gas, such as hydrogen, oxygen or carbon dioxide.

[0311] The reaction parameter may specify that the chemical material is only provided for part of the chemical synthesis. For example, the reaction parameter may specify that the chemical material is only added for the work up of a reaction product, or only for quenching a reaction. Such parameters may specify that the material is provided to the reaction space at some time period after the other materials have been combined, or after a reaction data provides a

[0312] 008889958certain output, thereby to quench a reaction or to permit the work up and possible isolation of reaction product.

[0313] The concentration of a material within a solution or in a suspension may be selected appropriately by the chemical synthesiser. The effective concentration of the material in the reaction space will depend on the concentration of that material within its individual chemical flow and the volume of other chemical materials with which it is combined in the reaction space. These volumes are dictated by the flow rates of each of the materials, which may be varied as appropriate, to alter the effective concentration of a material in the reaction space. Such techniques will be familiar to those with an understanding of flow chemistry techniques.

[0314] Where appropriate, a chemical material may be stored under an inert atmosphere, may be stored under anhydrous conditions or may be stored at reduced temperature, as required.

[0315] In one embodiment, there is provided 2 or more, 4 or more, 6 or more, or 10 or more chemical materials. For example, the flow chemistry system may comprise a number of controllable syringes equal to the number of specified chemical materials.

[0316] From time to time, it may be necessary to replenish a chemical material. The method of the invention need not be halted to allow such replenishment, and the chemical material may be replenished at such a time as it is not required during the chemical synthesis. The control system may be suitably programmed to predict the time at which a chemical material will become depleted. An operator may be warned accordingly. The control system may also be suitably programmed to factor into the decision making and control process the unavailability of an input owing to replenishment. The control system can continue to produce products using inputs other than the input that is being replenished.

[0317] The number of chemical reaction parameters may be one, though in this embodiment the number of physical reaction parameters, which may bring about a change in the chemical material, will typically be larger than the number of chemical reaction parameters.

[0318] In one embodiment, the instruction set may comprise two or more, three or more, four or more, five or more, six or more, ten or more, twenty or more chemical reaction parameters.

[0319] In one embodiment, the instruction set may comprise two or more, three or more, four or more, five or more, six or more, ten or more, twenty or more physical reaction parameters.

[0320] In preferred embodiments, the reaction parameter may be a reagent volume, reagent mass, reagent addition rate, solvent volume, solvent mass, solvent addition rate, reaction temperature, rate of change of reaction temperature, reaction pressure, rate of change of reaction pressure, pH, stirring rate, reaction mixture colour, reaction time, or a combination thereof.

[0321] 008889958In some embodiments, each of the reaction parameters in the instruction set are able to be adjusted, to provide an adjusted instruction set. Preferably, each of the reaction parameters in the instruction set are adjusted, to provide an adjusted instruction set.

[0322] In some embodiments, each of the reaction parameters in the instruction set are able to be optimised. Preferably, each of the reaction parameters in the instruction set are optimised.

[0323] Chemical Reactions

[0324] The method comprises performing chemical reactions in a chemical synthesiser. The chemical reactions are performed according to the initial instruction set and / or according to each of the derivative instruction sets. Each of the chemical reactions is performed in a different reaction vessel of the chemical synthesiser such that a reaction product is produced for each of the chemical reactions.

[0325] An instruction set is an executable description of a method for obtaining the reaction product by chemical reaction using a chemical synthesiser, such as an automated chemical synthesiser. The chemical reaction may be defined by the instruction set. Thus, the instruction set provides apparatus agnostic description of chemical operations needed to perform a chemical reaction. The reaction parameters of the instruction set define the reagents, conditions, and processing information that makes up the chemical synthesis. The instruction set is a typically a machine readable instruction set, coded using xDL.

[0326] One chemical reaction may be performed in parallel to a second chemical reaction. For example, at least two of the chemical reactions according to the number of derivative instruction sets, and optionally the initial instruction set, may be performed in parallel.

[0327] Preferably, each of the chemical reactions may be performed according to the number of derivative instruction sets, and optionally the initial instruction set, in parallel.

[0328] The number of chemical reactions to be performed is dependent on the number of derivative instruction sets generated, which itself is dependent on the complexity of the instruction set for the chemical reaction, and may also be dependent on the availability of the chemical synthesiser. Thus, from the point of view of time and resource efficiency, the number of chemical reactions to be performed to optimise the instruction set for the chemical reaction may be a minimum number of chemical reactions to be performed based on the complexity of the instruction set, and also the availability of the chemical synthesiser.

[0329] The number of derivative instruction sets to be generated and the number of chemical reactions to be performed may be a number based on the complexity and availability of the chemical synthesiser, as described above. This number may be a balance between reaction space exploration and time and resource efficiency.

[0330] 008889958For example, the number of derivative instruction sets to be generated and the number of chemical reactions to be performed may not be too high, where time and resources may otherwise be used in excess relative to complexity. The number of derivative instruction sets to be generated and the number of chemical reactions to be performed may also not be too low, where the complexity may require additional chemical reactions to be performed to explore chemical space.

[0331] Each of the chemical reactions is performed in a different reaction vessel of a chemical synthesiser. Thus, each chemical reaction produces its own reaction product. Each reaction product may have a different characteristic. At least one derivative instruction set may produce a reaction product with a desired characteristic.

[0332] Each of the chemical reactions may be performed before the predictive model is developed for optimisation by using the recorded reaction data and / or product data as described herein.

[0333] Reaction Data and Product Data

[0334] The method comprises a step of recording reaction data during the chemical synthesis and recording product data on the reaction product.

[0335] The reaction data is recorded during the chemical synthesis. The reaction data refers to data obtainable from the reaction mixture, reaction environment or surroundings. This refers to each test synthesis performed. The reaction data is recorded through all or some steps of the chemical synthesis.

[0336] The chemical synthesiser of the invention is provided with an analytical unit for recording reaction data during the chemical synthesis. The analytical unit then reports the reaction data to the control unit.

[0337] Reaction mixture may refer to the reaction mixture itself or the reaction environment.

[0338] Reaction environment may refer to the reaction mixture as held in a reaction vessel, and may include the headspace over the reaction mixture, where such is present. The reaction data may include data on the surroundings during the reaction. Surroundings refers to outside of the reaction vessel, for example the environment surrounding the chemical synthesiser (e.g., a fume hood).

[0339] The reaction data is data generated throughout the chemical synthesis. Thus, reaction data may be generated for each chemical reaction in the chemical synthesis. Additionally, the reaction data may be generated during the reaction itself, but may also be generated during the initial setting of the reaction mixture, and also the work-up of a reaction, including filtering or purification.

[0340] 008889958The reaction data is typically used in the step of adjusting one or more reaction parameters, as described herein. The use of reaction data in predicting how adjustments to the reaction parameter will improve the characteristic of the chemical reaction provides for more efficient and more effective optimisation of the chemical reaction.

[0341] The reaction data may be recorded at a set frequency, a variable frequency or continuously. The reaction data is preferably recorded at a high frequency, such as at time intervals of 1 second or less, 0.1 seconds or less, 0.01 seconds or less.

[0342] In some embodiments, the reaction data is recorded continuously throughout the chemical reaction.

[0343] In some embodiments, more than one different reaction data, such as two or three reaction data, are recorded simultaneously.

[0344] Reaction data may include physical or chemical properties. The reaction data may include reaction mixture colour, reaction mixture temperature, reaction mixture pressure, reaction mixture pH, reaction mixture filter rate, surrounding humidity, surrounding pressure, surrounding temperature, or a combination thereof, preferably reaction mixture colour, reaction mixture temperature or a combination thereof.

[0345] In some embodiments, the recording of reaction data changes depending on the reaction parameters. For example, where the reaction parameter includes the addition of a coloured reagent, the reaction data may include reaction mixture colour. However, where the reaction parameter includes the addition of only colourless reagents, the reaction data may not include reaction mixture colour.

[0346] In some embodiments, the recording of reaction data changes throughout the chemical synthesis. For example, where the chemical synthesis includes a heating step, the reaction data may include reaction mixture temperature. However, if the final work-up is carried out at ambient temperature, the reaction data may not include reaction temperature.

[0347] Optimised Instruction Set

[0348] The method comprises generating an optimised instruction set based on the predictive model. Specifically, the optimised instruction set is generated from the recorded reaction data and / or product data, which is used and analysed by the predictive model.

[0349] The optimised instruction set is an instruction set for a chemical reaction which, when the chemical reaction is performed, results in obtaining a desired characteristic of the chemical reaction. The desired characteristic may be a desired characteristic as described herein.

[0350] 008889958For example, the desired characteristic may relate to a reaction product having a desired characteristic.

[0351] The method may further comprise a step of performing a chemical reaction according to the optimised instruction set. The reaction data and / or the product data may be recorded for the chemical reaction performed according to the optimised instruction set, and may be compared against previous chemical reactions performed to validate the optimisation. This comparison may be used to validate if the optimised instruction set is better than the reactions performed during the optimisation process.

[0352] Chemical Synthesiser

[0353] In a second aspect of the invention there is provided a chemical synthesiser for optimising an instruction set for a chemical reaction, the chemical synthesiser comprising a plurality of reaction vessels, an analytical unit and a control unit, wherein:

[0354] the control unit is for obtaining an initial instruction set for a chemical reaction, and for receiving a number of derivative instruction sets generated from the initial instruction set using a predictive model, where each of the derivative instruction sets is different;

[0355] the plurality of reaction vessels is for performing chemical reactions according to the derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performable in a different reaction vessel to produce a reaction product for each of the chemical reactions; and

[0356] the analytical unit is for recording reaction data during each of the chemical reactions and / or recording product data on the reaction product of each of the chemical reactions; wherein the control system is for updating the predictive model using the reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction, and for generating an optimised instruction set based on the predictive model.

[0357] The present invention also provides for the chemical synthesiser, or parts of the chemical synthesiser, to operate autonomously. Thus, the chemical synthesiser may include an autonomous chemical synthesiser, suitably programmed to carry out the chemical synthesis according to the instruction set. The chemical synthesiser may include an autonomous analytical unit, suitably programmed to record reaction data and product data and report the data to a control system. The chemical synthesiser may include an autonomous control unit, suitably programmed to obtain an initial instruction set for a chemical reaction, to receive a number of derivative instruction sets generated from the initial instruction set, and transmit the reaction data and product data to a predictive model for optimisation of a characteristic of the chemical reaction, and enabling an optimised instruction set to the chemical synthesiser comprising one or more adjusted reaction parameters.

[0358] 008889958Suitably, the chemical synthesiser is for performing chemical reactions in each reaction vessel at laboratory scale. Such synthesis scales for each reaction vessel may correspond to a volume of between 0.1 to 30 mL.

[0359] In some cases, the reaction mixture may have a mass of 30 mL or less, such as 25 mL or less, such as 20 mL or less, such as 15 mL or less, such as 10 mL or less.

[0360] In some cases, the reaction mixture may have a mass of 0.1 mL or more, such as 1 mL or more, such as 2 mL or more, such as 3 mL or more, such as 4 mL or more.

[0361] The volume of the reaction vessel may be in an amount with the lower and upper volumes selected from the volumes given above. For example, the reaction mixture may have a mass of between 0.1 to 30 mL, such as 1 to 25 mL, such as 2 to 20 mL, such as 3 to 15 mL, such as 4 to 10 mL.

[0362] The present invention also provides a chemical synthesis platform comprising the chemical synthesiser of the second aspect, and one or more additional chemical synthesisers, wherein at least one of the additional chemical synthesisers is different to the chemical synthesiser.

[0363] The chemical synthesiser of the second aspect may be used as part of a chemical synthesis platform.

[0364] The method of the first aspect may be used as part of an optimisation campaign for optimising a series of reactions. The series of reactions may form part of a synthetic pathway.

[0365] In some embodiments, the complexity of the initial instruction set for each reaction in the synthetic pathway is determined based on the degree of structural difference between the starting material and the reaction product of the chemical reaction, the reaction type of the chemical reaction, the reaction mechanism of the chemical reaction and / or literature reports of the chemical reaction, where the chemical reaction is defined by the initial instruction set. The complexity may be determined as described herein. The complexity of each reaction may then be used to determine how the reaction is optimised on the chemical synthesis platform. An example of this is illustrated in Figure 2. For example, a high complexity reaction may be optimised using the chemical synthesiser of the second aspect, while a low complexity reaction may be optimised using an additional chemical synthesiser including only one reaction vessel.

[0366] In this way, the optimisation of a series of reactions can be assigned to the appropriate chemical synthesiser in the platform for the complexity of the step, such that the optimisation of the synthetic pathway may be achieved more efficiently.

[0367] 008889958Reaction Vessel

[0368] The chemical synthesiser comprises a plurality of reaction vessels for performing chemical reactions according to an instruction set. Each reaction vessel is for performing a chemical reaction according to the initial instruction set and / or each of the derivative instruction sets.

[0369] The chemical synthesiser is optionally also provided with one or more modules for work-up and purification, and further comprising reagents, optionally together with solvents and catalysts, for use in the chemical synthesis.

[0370] The chemical synthesiser may comprise a reactionware cartridge, for the performance of the chemical synthesis according to the instruction set. The reactionware may include multiple reaction vessels which are fluidically connected. The reactionware cartridge may be a reusable 3D printed ‘module-monolith’ reactionware cartridge. The reactionware cartridge is optionally also provided with one or more modules for work-up and purification, and further comprising reagents, optionally together with solvents and catalysts, for use in the chemical synthesis. The reactionware cartridge may be automatically generated from literature procedures using an intelligent software system based on the open-source universal chemical programming language standard and format, xDL.

[0371] The chemical synthesiser can be equipped with the components suitable for performing a chemical synthesis, preferably an automated chemical synthesis.

[0372] For example, the chemical synthesis may comprise a reagent and solvent handling system, such as a fluid handling system, a reaction vessel, a reagent / solvent supply system and waste system.

[0373] The chemical synthesiser typically comprises a fluid handling system. The fluid handling system may have any suitable form, and typically includes a network of pipes, valves, and pumps (e.g., syringe pumps) fluidically connected to the reaction vessel. The fluid handling system preferably includes a flow sensor, such as an optical flow sensor to measure the rate of flow of fluid into the reaction vessel.

[0374] The chemical synthesiser typically comprises a reaction vessel. The reaction vessel typically includes a reactor (e.g., a glass flask, flow cell, etc.), an agitator (e.g., sonicator or stirrer), a heater (e.g., a hot plate) and a cooler (e.g., a chiller).

[0375] The chemical synthesiser may also include a purification module, such as a column chromatography module, a filtration module, a washing module. The modules are suitable for performing their respective task. For example, the chromatographic module may be used for partial separation of reaction components for analysis, such as a low-pressure liquid

[0376] 008889958chromatography (LPLC) module, a high-pressure liquid chromatography (HPLC) module, an ultra-performance liquid chromatography (LIPLC) module, or a gas chromatography (GC) module.

[0377] The chemical synthesiser may also comprise a solid handling system, such as a solid reagent handling system. The solid handling system may be an automated solid handling system. For example, the solid handling system may be used for dispensing solids between modules of the chemical synthesiser, and for dispensing solids into and out of the reaction vessel.

[0378] The chemical synthesiser is preferably a robotic chemical synthesiser for autonomous performance of the chemical synthesis, and for autonomous control of the analytical device.

[0379] The function of the chemical synthesiser is as described in the performing chemical synthesis section, described above.

[0380] An analysis unit may be used in combination with a chromatographic unit for the at least partial separation of reaction components for analysis, such as an LPLC unit, an HPLC unit, a UPLC unit, or a GC unit.

[0381] Analytical Unit

[0382] The chemical synthesiser comprises an analytical unit is for recording reaction data from the reaction vessel during the chemical synthesis and for recording product data on the reaction product, and for reporting the reaction data and product data to the control unit.

[0383] The analytical unit comprises analytical devices. The analytical device includes sensors for measuring reaction data and product data.

[0384] Typically, the analytical unit comprises a reaction analysis unit for recording reaction data during the chemical synthesis and a product analysis unit for recording product data on the reaction product.

[0385] The analytical unit typically comprises a hub (referred to as a ‘sensor hub’) for centrally connecting the sensors (e.g., the sensors of the reaction analysis unit and the product analysis unit).

[0386] In some embodiments, the reaction analysis unit comprises a colour sensor, a temperature sensor, a pressure sensor, a pH sensor, a flow rate sensor, a humidity sensor or a combination thereof.

[0387] 008889958The temperature sensor, a pressure sensor, a humidity sensor may be a reaction mixture sensor (sensors configured to measure the reaction mixture) or surrounding sensors (sensor configured to measure the surroundings). For example, the reaction analysis unit comprises a reaction mixture colour sensor, a reaction mixture temperature sensor, a reaction mixture pressure sensor, a reaction mixture pH sensor, a reaction mixture flow rate sensor, a reaction mixture humidity sensor or a combination thereof. For example, the reaction analysis unit comprises a surrounding temperature sensor, a surrounding humidity sensor, a surrounding pressure sensor, or a combination thereof.

[0388] Preferably, the reaction analysis unit comprises a reaction mixture colour sensor, a reaction mixture temperature sensor or a combination thereof.

[0389] In some embodiments, the product analysis unit comprises an infra-red spectrometer, a Raman spectrometer, an NMR spectrometer, a mass spectrometer, a gas chromatography instrument, a high-performance liquid chromatography instrument, or a combination thereof. The NMR spectrometer may be used to perform1H NMR spectrometry or13C NMR spectrometry. The mass spectrometer may be selected from a quadrupole MS, a TOF-MS, an ion trap MS, an orbitrap MS, an LC-MS, an FT-ICR MS, a MALDI-TOF and an ESI-MS.

[0390] The function of the analytical unit is as described in the recording reaction data and product data steps, described above.

[0391] Control System

[0392] The chemical synthesiser comprises a control system for obtaining an initial instruction set for a chemical reaction, for generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different, for updating the predictive model using the reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction, and for generating an optimised instruction set based on the predictive model.

[0393] In some embodiments, the control system comprises a control unit, and the predictive model is located within a control unit of the chemical synthesiser. Thus, the control unit may be suitably programmed with a predictive model to generate a number of derivative instruction sets from the initial instruction set.

[0394] In some embodiments, the control system comprises a control unit and a server in communication with the control unit, and the predictive model is stored on the server.

[0395] In some embodiments, the server is for obtaining an initial instruction set for a chemical reaction and for generating a number of derivative instruction sets from the initial instruction

[0396] 008889958set using a predictive model, and the server sends the derivative instruction sets and optionally the initial instruction set to the control unit.

[0397] In some embodiments, the server is for receiving reaction data and / or product data of each of the chemical reactions from the control unit, and the server is for updating the predictive model using the reaction data and / or product data to optimise one or more characteristics of the chemical reaction, and for generating an optimised instruction set based on the predictive model, and optionally sending the optimised instruction set to the control unit.

[0398] The instruction sets, such as the initial instruction set and the derivative instruction sets, are typically stored in the control unit.

[0399] In some embodiments, the control unit is for analysing the reaction data and product data received from the analytical unit to determine a characteristic of the synthesis, wherein the characteristic of the synthesis is product yield, residual starting material amount, product purity, impurity amount, side-product amount, preferably product yield or residual starting material amount, more preferably product yield.

[0400] Preferably, the control unit receives reaction data during the chemical synthesis, and the control unit dynamically adjusts one or more dynamic reaction parameters during the chemical synthesis in response to the reaction data. Preferably, the control unit adjusts the dynamic reaction parameter such that the reaction data fulfils a different reaction parameter.

[0401] Thus, the control unit may be suitably programmed with an algorithm to analyse reaction data, product data and reaction parameters from previous test synthesis, and to adjust the reaction parameters for the next test synthesis.

[0402] The function of the control unit is as described in the analysing data and adjusting instruction set steps, described above.

[0403] Other Preferences

[0404] Each and every compatible combination of the embodiments described above is explicitly disclosed herein, as if each and every combination was individually and explicitly recited.

[0405] Various further aspects and embodiments of the present invention will be apparent to those skilled in the art in view of the present disclosure.

[0406] “and / or” where used herein is to be taken as specific disclosure of each of the two specified features or components with or without the other. For example “A and / or B” is to be taken as specific disclosure of each of (i) A, (ii) B and (iii) A and B, just as if each is set out individually herein.

[0407] 008889958Unless context dictates otherwise, the descriptions and definitions of the features set out above are not limited to any particular aspect or embodiment of the invention and apply equally to all aspects and embodiments which are described.

[0408] Certain aspects and embodiments of the invention will now be illustrated by way of example and with reference to the figures described above.

[0409] Examples

[0410] The following examples are provided solely to illustrate the present invention and are not intended to limit the scope of the invention, as described herein.

[0411] Example of General Parallel Optimisation

[0412] The complexity predictor estimates the likelihood of success of the next reaction attempt in the laboratory. While the output of the complexity predictor is a yield estimation (percentage from 0 to 100%), success is currently defined as an effective transformation according to a mass yield > 20%, but can be any arbitrary or objectively derived value. This assessment is based on multiple quantifiable factors that can be evaluated using automated systems.

[0413] In the context of the platform, serial predictions can be concentrated assuming negative results for each prior prediction to create a set of experiments spanning the chemical space quickly. In other words, where a set of conditions did not work, the platform makes a decision on a new set of conditions that should work.

[0414] Once some conditions are found, the yield prediction algorithm underlying the complexity predictor can be used to improve conditions towards a local (de minimis) or global (ideally, hard to prove) maxima through a “gradient descent” like process.

[0415] The analysis is structured into three main areas, each of which includes a reaction metric which corresponds to a particular complexity:

[0416] • Reaction classification and similarity to known reactions

[0417] • Reagent properties

[0418] • Substrate properties

[0419] Reaction Classification: Reactions are classified using a database containing 130 reaction SMARTS (SMILES arbitrary target specification) patterns. If a reaction matches a known class, it is assigned a value of 1 ; otherwise, it is assigned a value of 0. Additionally, reactions that fall within a subset of eight well-established and reliable reaction classes receive a separate classification value of 1. These elite classes include, for example:

[0420] • Suzuki coupling

[0421] 008889958• Amide coupling

[0422] • Reductive amination

[0423] • Carboxylic acid esterification

[0424] • Ester hydrolysis

[0425] • Sonogashira coupling

[0426] • Double bond reduction

[0427] Side Product Prediction: If a reaction is classified, its SMARTS pattern is used to assess the number of reactive centres and potential side products. For example, an amide coupling involving a molecule with two amine groups would be predicted to generate two products.

[0428] Reaction Similarity to known reactions (both successful and unsuccessful): For classified reactions, similarity is computed by comparing them to known reactions documented in literature (for example purchased by Pistachio database, set to low to medium trust) and internal data (high trust). The similarity score is determined using the cosine distance metric applied to reaction fingerprint vectors, identifying the most similar reaction in each database.

[0429] Reagent Properties: Reagents such as acids, bases, catalysts, and ligands can either be manually specified or automatically identified through a statistical analysis of the 200 most similar reactions in the Pistachio database. The most commonly used reagent set is selected for evaluation.

[0430] Inert Atmosphere Requirements: Using data from 7,100 reactions logged in an internal electronic laboratory notebook, correlations were established to determine which reagents require inert conditions. Out of 400 reagents analyzed, 53 were identified as requiring such conditions. If any reagents in a reaction fall into this category, the reaction is classified as requiring an inert atmosphere.

[0431] Hazard Assessment: Reagent hazard levels are determined by cross-referencing hazard phrases with an internal inventory database. If a reagent requires specialized handling or training, the reaction is flagged accordingly. Similar to inert conditions, the combined hazard level of all reagents contributes to the reaction assessment.

[0432] Substrate Complexity: A general complexity metric for substrates is calculated by counting the number of functional groups present. A predefined list of 36 SMARTS patterns corresponding to common organic functional groups is used, including:

[0433] • Alkynes, Alkenes, Arenes

[0434] • Aldehydes, Amides, Carboxylic acids, Cyanamides

[0435] • Esters, Ketones, Ethers, Enamines

[0436] • Primary, Secondary amines, Amino acids

[0437] • Azides, Azoles, Hydrazines, Imines

[0438] • Nitrates, Nitriles, Isonitriles, Nitros, Nitro groups

[0439] 008889958• Hydroxyls, Phenols, Phosphoric acids, Thiols

[0440] • Carbo-thioesters, Thioamides, Monosulfides, Disulfides

[0441] • Sulfones, Sulfoxides, Sulfates, Halides, Acyl halides

[0442] Generalizability: The approach has been generalized using 1) physical property calculations on the molecules and 2) fingerprinting techniques. List for 2 is in a separate document.

[0443] Currently the algorithms work in hybrid mode (using SMARTS classification when available to boost performance); though long term performance improvement will remove this additional classification requirements.

[0444] The total count of functional groups across all substrates is used as a complexity metric.

[0445] In summary, each reaction is evaluated based on the following eight key metrics:

[0446] Classified: Whether the reaction can be classified using the SMARTS database. Elite Class: Whether the reaction belongs to one of the eight elite classes.

[0447] Top Similarity (Chemify): Similarity score of the most similar internal reaction.

[0448] Top Similarity (Pistachio): Similarity score of the most similar external reaction.

[0449] Number of Products: Number of potential products.

[0450] Functional Group Count: Total number of functional groups in substrates.

[0451] Training Requirement: Whether any reagents require specialized handling.

[0452] Inert Atmosphere Requirement: Whether the reaction requires an inert atmosphere.

[0453] Figure 5 shows the correlation between the complexity values of 545 different types of chemical reactions, where the complexity is manually assigned by chemists (‘true label) and predicted by an XGBoost machine learning model (‘predicted label’), and where the correlation is visualised according to (top) a regression plot and (bottom) a classification plot.

[0454] The reaction metrics were calculated for 545 reactions within an internal database, corresponding to a commercial project where chemists manually assigned complexity values. This dataset was used to train an XGBoost machine learning model, with a 20% train-test split to validate model performance. Both classification and regression models were tested, demonstrating a correlation with manual complexity assessments.

[0455] Manual complexity assignments are inherently subjective, as the concept of reaction complexity involves numerous interdependent factors. Additionally, individual evaluators may weigh these factors differently. Validating the model against manually assigned complexity scores could propagate these biases.

[0456] An alternative approach would be to evaluate reaction complexity based on empirical success rates. By analysing internal experimental logs, we could determine how many times a given reaction was attempted under different conditions before achieving a satisfactory yield. This would provide a more objective measure of reaction difficulty.

[0457] 008889958Further cheminformatics metrics for substrates may also include:

[0458] Quantum Mechanical Calculations: Assessing the stability of transition states.

[0459] Steric Hindrance Analysis: Evaluating spatial constraints around reactive functional groups. Side Reaction Predictions: Identifying potential competing reactions.

[0460] Featurisation Techniques

[0461] A number of featurisation techniques are described below. These explain ‘fingerprints’ whereby physical features of atoms and molecules in reaction substrates are mapped onto computer-based descriptors.

[0462] 1. Structural (Bit-Based) Fingerprints

[0463] These fingerprints encode molecular structure as binary bit strings, where each bit represents the presence or absence of a specific molecular feature.

[0464] Morgan Fingerprints (ECFP, FCFP)

[0465] • Extended-Connectivity Fingerprints (ECFP) - Based on circular neighborhoods around atoms.

[0466] • Functional-Class Fingerprints (FCFP) - Similar to ECFP but considers functional groups.

[0467] Daylight-like Fingerprints

[0468] • Uses substructure hashing, where predefined molecular fragments are converted into bit vectors.

[0469] MACCS Keys (Molecular ACCess System)

[0470] • A widely used 166-bit fingerprint based on predefined chemical substructures.

[0471] Path-Based Fingerprints

[0472] • Encode molecular paths (e.g., linear, branched) up to a specific length.

[0473] Atom-Pair Fingerprints

[0474] • Encodes atom pairs and their topological distances.

[0475] Substructure Fingerprints

[0476] • Represent common chemical motifs like benzene rings, hydroxyl groups, etc.

[0477] Topological Torsion Fingerprints

[0478] • Based on four-atom torsion sequences in a molecule.

[0479] 2. Pharmacophore-Based Fingerprints

[0480] 008889958These fingerprints capture the 3D arrangement of functional groups relevant to biological activity.

[0481] Pharmacophore Fingerprints

[0482] • Represent molecular features like hydrogen bond donors / acceptors, charges, and hydrophobic centers.

[0483] SHED Fingerprints (SHannon Entropy Descriptors)

[0484] • Captures pharmacophore-related information in a compact format.

[0485] 3. Graph-Based Fingerprints

[0486] These fingerprints use molecular graphs to represent structures.

[0487] Graph Neural Network (GNN) Fingerprints

[0488] • Learn molecular embeddings from molecular graphs (used in deep learning).

[0489] Weisfeiler- Lehman Graph Kernel Fingerprints

[0490] • Encodes graph-based molecular similarity.

[0491] 4. Descriptor-Based Fingerprints

[0492] These fingerprints use physicochemical properties instead of structural features.

[0493] Molecular Property- Based Fingerprints

[0494] • Includes descriptors like molecular weight, logP, and hydrogen bond donors / acceptors.

[0495] Dragon Descriptors

[0496] • A large set of molecular descriptors used in cheminformatics.

[0497] RDKit Descriptors

[0498] • A set of computed molecular properties from the RDKit library.

[0499] PaDEL Descriptors

[0500] • A collection of molecular descriptors widely used in QSAR studies.

[0501] 5. Machine Learning & Deep Learning Fingerprints

[0502] These fingerprints leverage neural networks for feature extraction.

[0503] Autoencoder-Based Fingerprints

[0504] • Uses unsupervised learning to learn molecular representations.

[0505] Graph Convolutional Neural Networks (GCN)

[0506] 008889958• Generates molecular embeddings from graph data.

[0507] Transformer-Based Molecular Embeddings (e.g., ChemBERTa, MolBERT)

[0508] • Uses NLP-inspired transformers to generate fingerprints.

[0509] Variational Autoencoder (VAE) Fingerprints

[0510] • Learns continuous molecular representations.

[0511] Recurrent Neural Network (RNN) Fingerprints

[0512] • Encodes SMILES strings using sequence-based models.

[0513] 6. Hybrid Fingerprints

[0514] These combine multiple approaches to enhance performance.

[0515] Mixed Fingerprints (e.g., Morgan + MACCS)

[0516] • Combines structural and predefined fragment-based features.

[0517] Graph + Descriptor Fusion

[0518] • Merges graph embeddings with physicochemical properties.

[0519] DeepChem Molecular Representations

[0520] • Uses deep learning-based representations from the DeepChem library.

[0521] Parallel Optimisation of Van Leusen Oxazole Synthesis of 5-(4-Nitrophenyl)oxazole

[0522]

[0523] The Van Leusen oxazole reaction was selected as an example for the synthesis of heterocyclic building blocks, common in the medicinal chemistry literature. Two reaction parameters of the synthesis, product yield (%) and reaction time, were investigated by changing the conditions of the synthesis, such as the reaction temperature, the type of solvent and the type of base.

[0524] In the parallel optimisation campaign, the process was executed automatically and in parallel using a total of 18 parallel reactors. Each reactor contained a 1 :1 mixture of TosMIC, 4-nitrobenzaldehyde and 3 mL of solvent. The reaction mixture was stirred at a temperature of 25 to 75 °C for a total of 6 hours, where a first sample was taken after 3 hours for analysis, and a second sample was taken after 6 hours (a further 3 hours) for analysis.

[0525] 008889958For the parallel optimisation campaign, the following reaction parameters were set:

[0526] • Solvent type - ethanol, 1,1 -dichloroethane or toluene.

[0527] • Base (DBU, K2CO3) volume - variable within 1.0 - 2.0 eq.

[0528] • Reaction temperature - variable within 25.0 and 75.0 °C range.

[0529] The reaction product was characterised as follows:

[0530] 1H NMR (400 MHz, CDCL3): 57.92 (s, 1 H), 7.79 - 7.56 (m, 4H), 7.44 (s, 1 H).

[0531] 13C NMR (101 MHz, CDCI3): 6151.66, 149.87, 132.96, 131.80, 124.81, 124.33, 118.55, 112.10.

[0532] Parallel Optimisation of Van Leusen Oxazole Synthesis of 4-Oxazol-5-ylbenzonitrile

[0533]

[0534] A further Van Leusen oxazole reaction was performed in an analogous way to the parallel optimisation campaign described above. Here, each reactor contained a 1:1 mixture of TosMIC, 4-cyanobenzaldehyde and 3 mL of solvent. The reaction mixture was stirred at a temperature of 25 to 60 °C for a total of 6 hours, where a first sample was taken after 3 hours for analysis, and a second sample was taken after 6 hours (a further 3 hours) for analysis.

[0535] Iterative Optimisation of Van Leusen Oxazole Synthesis

[0536]

[0537] In the iterative optimisation campaign, the process was executed automatically using 2 parallel reactors. 0.25 M TosMIC in MeOH solution (4.10 - 6.15 mL), 0.25 M

[0538] 4-formylbenzonitrile in MeOH (4.1 mL, containing 0.05 M naphthalene as an internal standard), neat DBU (0.15 - 0.31 mL) and methanol (5 mL) were added to the reactor. The reaction mixture was stirred for 30 - 180 minutes at 25.0 - 75.0 °C. After cooling to room temperature, a sample is withdrawn from the reactor, 40 times diluted in an empty flask, and subsequently loaded onto a 5 mL sample loop and injected into the HPLC. The remaining

[0539] 008889958volume of the reaction mixture was discarded, and the platform reset by cleaning all modules with methanol and / or acetonitrile.

[0540] For the iterative closed-loop optimization the following reaction parameters were set:

[0541] • TosMIC volume - variable within 4.10 and 6.15 mL range (1.0 - 1.5 eq.)

[0542] • DBU volume - variable within 0.15 and 0.31 mL range (1.0 - 2.0 eq.)

[0543] • Reaction temperature - variable within 25.0 and 75.0 °C range.

[0544] • Reaction time - variable within 1800.0 and 10800.0 seconds.

[0545] The optimization target was set as the weighted sum of the peak purity and product peak area (14.5 min retention time) relative to the peak area of the internal standard (naphthalene, 20 min retention time), see Equation 1 where AUC refers to the area under the curve for a given peak, w is a weighing factor set to 0.7 and c is constant to scale the two objective values to the same order of magnitude set to 0.1. The optimization algorithm was SNOBFIT as implemented in the Summit framework.

[0546] (Equation 1)

[0547]

[0548] V 7

[0549] Configuration file to run Van Leusen oxazole synthesis using SNOBFIT algorithm via the Summit server framework with 2 parallel reactors:

[0550]

[0551] The reaction product was characterised as follows:

[0552] 1H NMR (400 MHz, CDCL3): 57.92 (s, 1 H), 7.79 - 7.56 (m, 4H), 7.44 (s, 1 H).

[0553] 13C NMR (101 MHz, CDCI3): 6151.66, 149.87, 132.96, 131.80, 124.81, 124.33, 118.55, 112.10.

[0554] Parallel Optimisation of Amide Coupling Reaction

[0555] 008889958

[0556]

[0557] As a further example of a parallel optimisation campaign, an amide coupling reaction (condensation of aniline and benzoic acid) was selected. The process was executed automatically and in parallel using a total of 24 parallel reactors. The reaction mixture was stirred at a temperature of 25 to 40 °C for a total of 2 hours.

[0558] For the parallel optimisation campaign, the following reaction parameters were set:

[0559] • Aniline amount - variable within 0.84 to 1.05 mmol relative to 0.7 mmol of benzoic acid

[0560] • Reagent volume - Reagent 2 (N-ethyl-N-isopropylpropan-2-amine) variable within 0.91 to 1.05 mmol relative to 1.05 mmol of Reagent 1 (2,4,6-tripropyl-1 , 3, 5, 2,4,6- trioxatriphosphinane 2,4,6-trioxide).

[0561] • Solvent volume - Solvent 1 (N,N-dimethylformamide) variable within 0 to 5 mL.

[0562] • Reaction temperature - variable within 25.0 and 40.0 °C range.

[0563] Optimisation Results and Discussion

[0564] The results from the two parallel optimisation campaigns for the Van Leusen Oxazole Synthesis for each reactor are shown in Table 1 and Table 3. For comparison, the results from the iterative optimisation campaign for each iteration are shown in Table 2.

[0565] In the first parallel optimisation campaign, the starting material is 4-nitrobenzaldehyde and the reaction product is 5-(4-nitrophenyl)oxazole. In the iterative optimisation campaign, the starting material is 4-formylbenzonitrile and the reaction product is 4-oxazol-5-ylbenzonitrile. A reasonable comparison may be made between the two optimisation campaigns due to the extensive structural similarity between the two starting materials and reaction products, as well as the similar reaction pathway.

[0566] In the second parallel optimisation campaign, the starting material is 4-cyanobenzaldehyde and the reaction product is 4-oxazol-5-ylbenzonitrile. The starting material and reaction product of the second parallel optimisation campaign is the same as that used for the

[0567] 008889958iterative optimisation campaign, and so a direct comparison may be made between the two optimisation campaigns.

[0568] The parallel optimisation campaign was concluded following the completion of 18 parallel reactions. As shown in Table 1 , the reaction with the highest yield after 6 hours was performed in reactor 13, and the yield was at 68.45%. From converting the yield of 68.45% to a value corresponding to a relative peak area, it was estimated that the relative peak area after 6 hours was 15.40.

[0569] As shown in Table 3, the reaction with the highest yield after 6 hours was performed in reactor 9, and the yield was at 90.59%. From converting the yield of 90.59% to a value corresponding to a relative peak area, it was estimated that the relative peak area after 6 hours was 20.4.

[0570] For comparison, as shown in Table 2, the reaction in the iterative optimisation campaign with the highest relative peak area was performed in iteration 2, and the relative peak area was 12.09. Accordingly, the first parallel optimisation campaign achieved a 27% increase in the peak area relative to the results from the iterative optimisation campaign. Furthermore, the second parallel optimisation campaign, which is identical in its starting material and reaction product to the iterative optimisation campaign, achieved a 69% increase in the peak area relative to the results from the iterative optimisation campaign.

[0571] Furthermore, the parallel optimisation campaign was concluded in 6 hours or less, in both cases. For comparison, each reaction in the iterative optimisation campaign was performed for a duration from 30 minutes to 3 hours, with an average iteration duration of over 30 minutes. The total time required for the iterative optimisation campaign, which required a total of 26 successive iterative reactions, was around 34 hours of machine time. Importantly, each successive iteration required a user to set up the reaction, thus requiring 26 touch points for the user during the optimisation. In contrast, the parallel optimisation was carried out by performing all the pre-determined reactions at the same time, meaning each reaction could be set up in one go by a user, thus requiring a single touch point for the user to achieve the optimisation.

[0572] The results from the parallel optimisation campaign for the Amide Coupling Reaction for each reactor are shown in Table 4. In this parallel optimisation campaign, the starting materials are aniline and benzoic acid, and the reaction product is benzanilide. A total of 24 reactions were performed, based on the complexity of this specific amide coupling reaction and the availability of the chemical synthesiser.

[0573] In conclusion, the results from the optimisation campaigns showed that the parallel optimisation achieved an improved yield in reaction product, while the time and resources required to complete the reaction optimisation campaign were also minimised.

[0574] 008889958Table 1: Results of the Van Leusen Oxazole parallel optimisation of 5-(4-Nitrophenyl)oxazole

[0575]

[0576] 008889958Table 2: Results of the Van Leusen Oxazole iterative optimization of 4-Oxazol-5-ylbenzonitrile

[0577]

[0578] 008889958Table 3: Results of the Van Leusen Oxazole parallel optimisation of 4-Oxazol-5-ylbenzonitrile

[0579]

[0580] 008889958Table 4: Results of the Amide Coupling Reaction parallel optimisation

[0581]

[0582] 008889958References

[0583] A number of publications are cited above in order to more fully describe and disclose the invention and the state of the art to which the invention pertains. Full citations for these references are provided below. The entirety of each of these references is incorporated herein.

[0584] WO 2024 / 091573

[0585] US 2021 / 125060

[0586] WO 2023 / 131726

[0587] US 5463564

[0588] US 2020 / 0225251

[0589] Du et al. Chemometr. Intell. Lab. Syst. 1999, 48, 181-203

[0590] Salley et al. Nat. Commun. 2020, 11, 2771

[0591] He et al. J. Am. Chem. Soc. 2024, 146, 28952-28960

[0592] 008889958

Claims

Claims:

1. A method of optimising an instruction set for a chemical reaction, the method comprising the steps of:(i) obtaining an initial instruction set for a chemical reaction;(ii) generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different;(iii) performing chemical reactions according to the number of derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performed in a different reaction vessel of a chemical synthesiser;(iv) recording reaction data during each of the chemical reactions and / or recording product data on a reaction product of each of the chemical reactions using an analytical unit of the chemical synthesiser;(v) updating the predictive model using the recorded reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction; and(vi) generating an optimised instruction set based on the updated predictive model.

2. The method of claim 1 , further comprisingdetermining the complexity of the initial instruction set using the predictive model, and calculating the number of derivative instruction sets to be generated based on the complexity of the initial instruction set.

3. The method of claim 2, wherein the complexity of the initial instruction set is based on a probability that performing a chemical reaction according to the initial instruction set will result in one or more desired characteristics of the chemical reaction, such as a desired characteristic of the reaction product.

4. The method of claim 2 or 3, wherein the complexity of the initial instruction set is compared to a threshold, and where the complexity is at or above the threshold the performing step is carried out on a chemical synthesiser having multiple reaction vessels, and where the complexity is below a threshold the performing step is carried out on a chemical synthesiser with a single reaction vessel.

5. The method of any one of claims 2 to 4, wherein the complexity of the initial instruction set is based on:(a) the degree of structural difference between the starting material and the reaction product of the chemical reaction;(b) the reaction type of the chemical reaction;(c) the reaction mechanism of the chemical reaction; and / or(d) literature reports of the chemical reaction.0088899586. The method of any one of claims 1 to 5, wherein steps (ii) to (v) are repeated based on the updated predictive model, wherein step (ii) comprises generating a number of derivative instruction sets from the updated predictive model.

7. The method of claim 6, wherein steps (ii) to (v) are repeated until the one or more characteristics of the chemical reaction are optimised.

8. The method of any one of claims 1 to 7, wherein each of the derivative instruction sets is different to the initial instruction set.

9. The method of any one of claims 1 to 8, wherein each of the chemical reactions is performed according to the each of the derivative instruction sets, and optionally the initial instruction set, before the predictive model is updated.

10. The method of any one of claims 1 to 9, wherein the initial instruction set is obtained from one or more literature synthetic procedures, such as wherein the initial instruction is a literature synthetic procedure.

11. The method of any one of claims 1 to 9, wherein the initial instruction set is obtained by modifying one or more literature synthetic procedures, optionally wherein the one or more literature synthetic procedures are for a different chemical reaction to the initial instruction set.

12. The method of any one of claims 1 to 11, wherein the reaction data is recorded throughout the chemical reaction.

13. The method of any one of claims 1 to 12, wherein at least two of the chemical reactions performed according to the number of derivative instruction sets, and optionally the initial instruction set, are performed in parallel; preferably wherein all the chemical reactions performed according to the number of derivative instruction sets, and optionally the initial instruction set, are performed in parallel.

14. The method of any one of claims 1 to 13, wherein step (vi) further comprises performing a chemical reaction according to the optimised instruction set.

15. The method of any one of claims 1 to 14, wherein the method is performed autonomously.

16. The method of any one of claims 1 to 15, wherein:(a) the one or more characteristics of the chemical reaction are reaction product yield, reaction product purity, reaction product to starting material ratio, reaction product to impurity ratio, solvent mass consumption, reagent mass consumption, reaction duration or a008889958combination thereof; preferably wherein the one or more characteristics is reaction product yield, and the yield is determined by NMR spectroscopy, Raman spectroscopy or high-performance liquid chromatography product data; and / or(b) the reaction data is reaction colour, reaction temperature, reaction vessel pressure, reaction pH, reaction filter rate, surrounding humidity, surrounding pressure, surrounding temperature, or a combination thereof, preferably reaction colour, reaction temperature, reaction pressure or a combination thereof; and / or(c) the product data is spectroscopic data, such as UV-Vis, Infra-red, Raman, NMR, or mass spectrometry data; chromatography data, such as gas or high-performance liquid chromatography data; or a combination thereof; and / or(d) the instruction set comprises reaction parameters selected from reagent volume, reagent mass, reagent addition rate, solvent volume, solvent mass, solvent addition rate, reaction temperature, rate of change of reaction temperature, reaction pressure, rate of change of reaction pressure, pH, stirring rate, reaction mixture colour and reaction time.

17. A chemical synthesiser for optimising an instruction set for a chemical reaction, the chemical synthesiser comprising a plurality of reaction vessels, an analytical unit and a control system, wherein:the control system is for obtaining an initial instruction set for a chemical reaction, and for generating a number of derivative instruction sets from the initial instruction set using a predictive model, where each of the derivative instruction sets is different;the plurality of reaction vessels is for performing chemical reactions according to the number of derivative instruction sets and optionally the initial instruction set, wherein each of the chemical reactions is performable in a different reaction vessel to produce a reaction product for each of the chemical reactions; andthe analytical unit is for recording reaction data during each of the chemical reactions and / or recording product data on the reaction product of each of the chemical reactions; wherein the control system is for updating the predictive model using the reaction data and / or product data of each of the chemical reactions to optimise one or more characteristics of the chemical reaction, and for generating an optimised instruction set based on the predictive model.

18. The chemical synthesiser of claim 17, wherein the chemical synthesiser is for autonomously optimising the instruction set.

19. The chemical synthesiser of claim 17 or 18, wherein the control system comprises a control unit and a server in communication with the control unit, and preferably the predictive model and / or the instruction sets are stored on the server.

20. The chemical synthesiser of any one of claims 17 to 19, wherein the analytical unit comprises a colour sensor, a temperature sensor, a pressure sensor, a pH sensor, a flow rate sensor, a humidity sensor or a combination thereof.00888995821. The chemical synthesiser of any one of claims 17 to 20, wherein the analytical unit comprises an infra-red spectrometer, a Raman spectrometer, a NMR spectrometer, a mass spectrometer, a gas chromatography instrument, a high-performance liquid chromatography instrument or a combination thereof.

22. The chemical synthesiser of any one of claims 17 to 21, further comprising a bottom module and a top module, wherein:the bottom module is for receiving the outside of the reaction vessel and is for applying reaction conditions of the instruction set to the outside of the reaction vessel, and the top module is for accessing the inside of the reaction vessel and for applying reaction conditions of the instruction set to the inside of the reaction vessel.

23. The chemical synthesiser of claim 22, wherein:(a) the bottom module is for receiving two or more reaction vessels, such as four or more reaction vessels, such as six or more reaction vessels; and / or(b) the top module is for accessing the inside of one reaction vessel.

24. A chemical synthesis platform comprising the chemical synthesiser of any one of claims 17 to 23 and one or more additional chemical synthesisers, wherein at least one of the additional chemical synthesisers is different to the chemical synthesiser, and preferably at least one of the additional chemical synthesisers has one reaction vessel.

25. Use of the chemical synthesiser of any one of claims 17 to 23, or the chemical synthesis platform of claim 24, for optimising an instruction set for a chemical reaction.008889958

Images