Controlling a system for assaying compounds

Abstract

The invention relates to controlling a system for assaying compounds. For each assay in an assay cascade, determining whether the respective assay is one of a plurality of defined assays stored in an assay database having a standard operating procedure, SOP, for performing the respective defined assay. If the respective assay is not one of the plurality of defined assays, then the invention involves determining an SOP including a value for each of a plurality of assay conditions for performing the respective assay. This involves comprises retrieving a plurality of defined SOP procedural rules for determining SOPs, and processing the plurality of defined SOP procedural rules to generate the SOP for performing the respective assay. The system is controlled to perform the respective assay on a plurality of synthesised compounds according to the SOP of the respective assay, and results of the respective assay are obtained from the system.

Description

[0001] CONTROLLING A SYSTEM FOR ASSAYING COMPOUNDS

[0002] TECHNICAL FIELD

[0003] The present invention relates to controlling a system for assaying compounds. In particular, the invention relates to automatically determining a standard operating procedure for one or more assays in an assay cascade, and automatically controlling the system to implement the assay cascade. Specifically, the standard operating procedure for the one or more assays is determined using rule-based procedural generation.

[0004] BACKGROUND

[0005] Drug discovery is the process of identifying candidate compounds / molecules for progression to the next stage of drug development, e.g. pre-clinical trials. Such candidate compounds are required to satisfy certain criteria for further development. Modern drug discovery involves the identification and optimisation of initial screening ‘hit’ compounds. In particular, such compounds need to be optimised relative to required criteria, which can include the optimisation of a number of different properties. The properties to be optimised can include, for instance: efficacy / potency against a desired target; selectivity against nondesired targets; low probability of toxicity; and, good drug metabolism and pharmacokinetic properties (ADME). Only compounds satisfying the specified requirements become candidate compounds that can continue to the drug development process.

[0006] The identification and progression of (small) molecules from early proof of concept to clinical candidate may typically be performed according to a so-called DMTL loop (‘design- make-test-learn’ loop). A DTML loop is a well-established approach for enhancing compound properties and driving molecules towards clinical candidates. It will be apparent that factors that reduce cycle times, or improve efficiency and accuracy, of iterative DMTL loops result in beneficial and cumulative impacts on the overall progression of drug discovery projects.

[0007] Recent years have seen a great deal of attention on increasing the efficiency of the ‘design’ phase of the DMTL loop. For instance, many artificial intelligence (Al) / machine learning (ML) methods have been developed to identify / design compounds predicted to exhibit various desirable properties, e.g. according to a desired property profile of a drug to be designed, so that the number of design cycles that need to be performed to identify hit compounds is reduced.

[0008] There has been less attention on the other phases of the DMTL loop. Many processes in the ‘make’ and ‘test’ loops, for instance, are still performed manually, or at least require manual intervention at certain points, e.g. to design or trigger the processes. In the ‘test’ phase, for instance, while it is known to automate steps associated with compound screening, other tasks require significant human intervention. For a given project, many different assays may be needed to test various aspects / properties of compounds, and it can typically take several months for these to be designed and implemented manually.

[0009] Also, most automation software for controlling equipment, e.g. robots, to perform tasks at one or more phases of the DMTL loop requires operators to program / instruct each step individually that is to be performed, e.g. ‘move pipette to position A7’. This is time consuming and prone to human error, not least because scheduling / programming software for this purpose is complex and requires specialist human input.

[0010] It is against this background to which the present invention is set.

[0011] SUMMARY OF THE INVENTION

[0012] The present invention provides systems and methods for increasing the efficiency of performing aspects of the ‘test’ phase of a DMTL loop (‘design-make-test-learn’ loop) as part of a drug discovery process. In particular, the invention provides systems and methods for automating aspects of the test phase in order to increase efficiency (throughput and speed), while adhering to constraints of predictability and reliability required in the test phase. The systems and methods of the invention also allow for complexity inherent in various tasks of the test phase to be handled and embraced, while keeping the systems / methods accessible for operators to monitor. To achieve these benefits, automation of the development and / or implementation of assays for assaying compounds of interest is provided via the systems / methods of the invention. That is, the invention provides for the automated design of experiments / assays. Specifically, this is achieved via the use of rules-based procedural generation to determine the various conditions / requirements for an assay based on the outcome to be achieved, e.g. enzyme titration. The invention furthermore optionally provides for the automated implementation of the designed / developed assays to assay compounds by using rules-based procedural generation to determine instructions for controlling equipment, e.g. robots, required to perform an assay, without needing human input to provide individual instructions.

[0013] According to an aspect of the present invention there is provided a computer-implemented method of controlling a system for assaying compounds. The method may comprise defining an assay cascade comprising a plurality of assays to be performed in a defined sequence. The method may comprise receiving an indication that a plurality of synthesised compounds are to be assayed according to the defined assay cascade. The method may comprise identifying a first assay, in the defined sequence, of the assay cascade. The method may comprise accessing an assay database storing a plurality of defined assays each including a standard operating procedure, SOP, for performing the respective defined assay. Each SOP may include a value for each of a plurality of assay conditions for performing the respective defined assay. The method may comprise determining whether the first assay is one of the plurality of defined assays stored in the assay database. If the first assay is one of the plurality of defined assays, then the method may comprise retrieving the SOP including the values of the plurality of assay conditions of said one of the defined assays from the assay database. If the first assay is not one of the plurality of defined assays, then the method may comprise determining an SOP including a value for each of the plurality of assay conditions for performing the first assay. The method may comprise controlling the system to perform the first assay on the plurality of synthesised compounds according to the SOP of the first assay, and may comprise obtaining one or more results of the first assay from the system.

[0014] Determining the SOP for performing the first assay may comprise: retrieving a plurality of defined SOP procedural rules for determining SOPs; and processing the plurality of defined SOP procedural rules to generate the SOP for performing the first assay.

[0015] The method may further comprise repeating steps of: identifying a next assay, in the defined sequence, of the assay cascade; determining a set of compounds, from the plurality of synthesised compounds, on which the next assay is to be performed, wherein the set of compounds is determined based on the one or more results of the previously- performed assay obtained from the system; accessing the assay database, and determining whether the next assay is one of the plurality of defined assays stored in the assay database; if the next assay is one of the plurality of defined assays, then retrieving the SOP including the values of the plurality of assay conditions of said one of the defined assays from the assay database; if the next assay is not one of the plurality of defined assays, then determining an SOP including a value for each of the plurality of assay conditions for performing the next assay, wherein determining the SOP for performing the next assay comprises retrieving the plurality of defined SOP procedural rules for determining SOPs and processing the plurality of defined SOP procedural rules to generate the SOP for performing the next assay; controlling the system to perform the next assay on the set of compounds according to the SOP of the next assay, and obtaining one or more results of the next assay from the system, until each of the assays in the assay cascade has been performed.

[0016] The plurality of defined SOP procedural rules may be processed based on one or more defined initial parameters. These may include one or more variables that can be adjusted and a range of values within which each respective variable can be adjusted.

[0017] The variables that can be adjusted may include one or more of: a concentration of a protein to be used in the assay; a concentration of a substrate to be used in the assay; and a concentration of a reference compound to be used in the assay.

[0018] The plurality of assays may include one or more of: an enzyme titration assay; and a binding assay.

[0019] The plurality of assay conditions may include one or more biochemical conditions. Optionally, the biochemical conditions include one or more of: buffer scouting; substrate optimal concentration; ideal time point; ideal target concentration; and dimethyl sulfoxide concentration.

[0020] The plurality of assay conditions may include one or more cell assay conditions. Optionally, the cell assay conditions include one or more of: ideal transfection conditions; and cell number.

[0021] The plurality of assay conditions may include one or more validation conditions for validating assay performance. Optionally, the validation conditions include one or more of: plate uniformity; and structure activity relationship sets.

[0022] Processing the plurality of defined SOP procedural rules to generate the SOP for performing the first assay may be performed as part of an iterative process. The iterative process may comprise: processing the plurality of defined SOP procedural rules to generate a candidate SOP including a candidate value for each of the plurality of assay conditions for performing the first assay; controlling the system to perform the first assay on one or more reference compounds according to the candidate SOP of the first assay, and obtaining one or more intermediate results of the first assay from the system; determining a value for each of one or more parameters each indicative of effectiveness of a respective aspect of the first assay, wherein the parameter values are determined based on the obtained intermediate results; and determining whether each determined parameter value satisfies a respective desired parameter condition.

[0023] If the determined parameter values satisfy the respective desired parameter conditions, then the candidate SOP may be determined to be the SOP for performing the first assay. If one or more of the determined parameter values do not satisfy the respective desired parameter conditions, then the method may comprise repeating steps of: processing the plurality of defined SOP procedural rules, based on the parameter values obtained from performing the first assay according to the previous candidate SOP, to generate an updated candidate SOP including an updated candidate value for each of the plurality of assay conditions for performing the first assay; controlling the system to perform the first assay on one or more reference compounds according to the updated candidate SOP of the first assay, and obtaining one or more intermediate results of the first assay from the system; and determining, based on the obtained intermediate results, the parameter values indicative of effectiveness of respective aspects of the first assay, until a stop condition is satisfied. The updated candidate SOP when the stop condition is satisfied may be determined to be the SOP for performing the first assay.

[0024] The reference compounds may include the plurality of compounds on which the first assay is to be performed.

[0025] The stop condition may be one of: the repeated steps have been performed a defined number of times; and a rate of improvement of the determined parameter values falls below a defined threshold rate of improvement.

[0026] The parameters indicative of assay effectiveness may include one or more of: a signal-to- noise ratio parameter; a surface response parameter; a Michaelis-Menten kinetics parameter; a Z factor parameter; a plate uniformity parameter; and a dose response uniformity parameter. Controlling the system to perform the first assay on the plurality of synthesised compounds according to the SOP of the first assay may comprise: retrieving a plurality of defined system steps procedural rules for determining process steps to be implemented by the system to perform assays; processing the plurality of defined system steps procedural rules, based on the SOP of the first assay, to generate the process steps for performing the first assay; and instructing the system to implement the generated process steps to perform the first assay.

[0027] Controlling the system to perform the next assay on the determined set of compounds according to the SOP of the next assay may comprise: retrieving the plurality of defined system steps procedural rules, and processing the plurality of defined system steps procedural rules, based on the SOP of the next assay, to generate the process steps for performing the next assay; and instructing the system to implement the generated process steps to perform the next assay.

[0028] Controlling the system to perform the first or next assay may comprise controlling one or more robotic devices of the system to retrieve and prepare materials needed to perform the first or next assay.

[0029] According to another aspect of the present invention there is provided a non-transitory, computer readable storage medium storing instructions thereon that, when executed by one or more computer processors, cause the one or more computer processors to perform the method defined above.

[0030] According to another aspect of the present invention there is provided a controller comprising one or more computer processors, the controller being configured to perform a method as defined above.

[0031] According to another aspect of the present invention there is provided a system for assaying compounds. The system comprises a controller as defined above. The system comprises one or more robotic devices configured to receive scheduling instructions from the controller to perform one or more actions to implement each assay in the assay cascade.

[0032] BRIEF DESCRIPTION OF THE DRAWINGS Examples of the invention will be described with reference to the accompanying drawings, in which:

[0033] Figure 1 schematically illustrates a flow diagram including steps for testing compounds, including assaying compounds, performed during a drug discovery process;

[0034] Figure 2 shows the steps of a method in accordance with an aspect of the invention;

[0035] Figure 3 shows procedural generation process for developing a substrate titration step of an assay in accordance with the method of Figure 2;

[0036] Figures 4 and 5 outline definitions for respective first and second models used in the process of Figure 3;

[0037] Figure 6 outlines a definition of a Michaelis Menten model used in the process of Figure 3;

[0038] Figure 7 shows procedural generation process for developing a probe titration and time course step of an assay in accordance with the method of Figure 2;

[0039] Figure 8 outlines a definition of a model used in the process of Figure 7;

[0040] Figure 9 shows procedural generation process for developing an antibody / probe versus protein titration step of an assay in accordance with the method of Figure 2; and

[0041] Figure 10 shows procedural generation process for developing a recommended DMSO tolerance of an assay in accordance with the method of Figure 2.

[0042] DETAILED DESCRIPTION

[0043] Examples of the invention provide computer-implemented systems and methods for automatically developing one or more assays - and, specifically, the conditions or parameter values needed to perform the assays - in a cascade or series of assays. As is well known in the art, an assay is a procedure or examination of a substance in a laboratory setting to assess or measure the presence, amount or activity, etc., of a target compound / molecule. Compounds may be assayed as part of a DMTL loop (‘design-make- test-learn’ loop) during a drug discovery project. In particular, compounds that have been synthesised at a ‘make’ phase of the DMTL loop may then be assayed as part of a ‘test’ stage.

[0044] Typically, a number of assays may be performed in sequence at the test stage, with only some of a batch of compounds being selected to proceed to a next assay in the sequence based on the results of a previous assay in the sequence. The assays in an assay sequence / cascade may generally become progressively more complex and / or expensive to perform through the cascade. Determining the conditions or parameter values needed to perform assays in a sequence may therefore become progressively more difficult and time consuming. For instance, manual design of assays for a given drug discovery project may take several months.

[0045] Figure 1 schematically illustrates a typical process 10 for developing and undertaking assaying of compounds, e.g. in a test phase of a project. At step 101 , the type of assay to be performed is defined. This could be an enzyme titration assay, binding assay, etc. Typically, a store 102 of defined reagents may be available to perform the assay(s). Based on the defined assay to be performed and the library of available reagents, the assay is designed at step 103. In particular, this includes the determining / designing the conditions for performing the assay, e.g. temperature, pH, concentrations of reaction components, ionic strength, buffer selection, etc. This typically involves defining a standard operating procedure (SOP) including step-by-step instructions for performing the assay.

[0046] At step 104, scheduling instructions for performing the assay are determined / defined. In particular, this involves programming, in software, instructions for automatically controlling the various pieces of laboratory equipment needed to perform the assay, e.g. liquid handling equipment, sample storage units, etc. At step 105, the assay is performed via the scheduling software controlling the laboratory equipment according to the defined scheduling instructions. At step 106, results of the assay are obtained / collected, e.g. measuring the amount of substance in a sample, and then at step 107 analysis of the results is performed, e.g. using computational techniques, and used as feedback to step 101 of defining the (next / updated) assay. The output of the analysis at step 107 may also be subject to user analysis, e.g. by inspection, at step 108, followed by a user design, step at step 109, to (also) feed back into the step of designing the assay (step 101 ).

[0047] It is required that predictable and reliable performance is achieved at each step of the process. Many steps of the process / cycle 10 are typically performed manually. For instance, steps 101 -103 may be regarded as an iterative process - rather than a design loop - when performed manually. Indeed, a human performing these steps needs to decide the process / assay to be performed and then determine what the outcome needs to be. As an example, the desired concentrations of samples needs to be determined. The human then needs to perform back-calculation to determine values such as final dilutions of samples based on volume. The human then further needs to determine initial dilutions based on final dilutions and available stock, also via back-calculation.

[0048] To perform step 104 manually, a human operator / user needs to program instructions on a ‘per step’ basis rather than at a workflow level. Indeed, a human will typically take the SOP defined at step 103 and program the scheduling software - for controlling the laboratory equipment - directly by programming each individual step of the SOP. To perform step 107 manually, a human operator / user may generate their own analysis protocol based on the SOP.

[0049] In examples of the invention, one or more steps of the process 10 are automated, i.e. performed automatically, rather than manually. For instance, automation of steps 101 -103 involves performing updates / determinations based on the stage of development (e.g. of an assay) and existing information (e.g. results of a previous design loop, what exists in the store / library 102, etc.). In particular, acquired information is fed into a rules-based procedural generator to determine (updated) assay conditions in a deterministic (predictable) manner, i.e. the output of a previous step is used to inform the next step in a deterministic manner.

[0050] Procedural generation allows for new content (instructions, experiment steps, experiment conditions, etc.) to be designed / obtained from a set of variable parameters and a rulebased level designer. This allows for a large degree of complexity without needing to define and store each possible variation of experimental parameters separately as stored ‘recipes’ of individual steps / conditions. This is highly beneficial as the space of possible combinations of experimental steps, conditions, etc., is typically vast. Procedural generation also allows for the predictable generation of new content from simple natural language prompts.

[0051] The automated approach allows for the steps to achieve a desired outcome to be determined automatically. For instance, the desired outcome may be to build an enzyme assay. In examples of the invention, the steps of an experiment to be performed (including the conditions to run the experiment) are determined automatically in a deterministic manner.

[0052] In examples of the invention, the step of scheduling instructions (step 104) is performed automatically. In particular, information from previous steps, e.g. the SOP, is provided to a rules-based procedural generator to determine the scheduling instructions. In particular, in such examples a human operator / user does not need to define the assay steps and the scheduling instructions for the entire process can be written / defined as a single step rather than a human defining each individual step to be undertaken to perform the assay. In these examples of the invention, no (further) human intervention is needed from the initial specification of what is desired (e.g. to perform an enzyme assay) in order for the assay to be both designed and then scheduled / performed, i.e. in these examples both of these are determined automatically using procedural generation. Furthermore, automated analysis at step 107 may be performed with information from previous steps being used to define and performed the analysis, with results then being presented to a user. Indeed, the outputs of the data collection and analysis steps 106, 107 may be used to inform the next step as part of a feedback loop without requiring human intervention.

[0053] Referring back to step 104 in which an assay is performed / run, consider a scenario in which a screen to be performed using a defined assay with known / defined parameters / conditions is triggered. This may involve an operator inputting a request to suitable scheduling software via suitable computer hardware (graphical user interface) to run a specific assay - e.g. assay X - on specific compounds - e.g. compounds Y. An example of scheduling software is Cellario™. The known assay X may be associated in the scheduling software with a specific assay ID (identification number / string) and the scheduling software can then retrieve the known / defined conditions for performing assay X from a database storing details for performing a plurality of defined assays.

[0054] The retrieved conditions / parameters are used to generate instructions for generating material plates needed to perform the assay. This includes generating instructions for reagent plate generation using reagent IDs retrieved from the database for assay X and compound plate generation using compounds IDs corresponding to the compounds selected to be assayed. The generated instructions include material transfer instructions for controlling equipment to move the materials to be used from material stores to relevant plates. In particular, automated material preparation for performing an assay may involve automated steps of sample selection, sample retrieval based on material IDs (from automated stores), and dispensing based on material transfer instructions (using automated liquid handling equipment). The generated reagent and compound plates are then combined. The prepared materials are then used to run assay X as instructed by the scheduling software. In particular, this will typically involve steps of incubation, readout and automatic analysis. The results can then be reported / returned to the operator, e.g. via the GUI.

[0055] Referring again to step 104 in which an assay is performed / run, consider a scenario in which a screen to be performed using an assay with unknown / undefined parameters / conditions is triggered. In this case, the operator may define an aim, outcome or goal of the assay and provide initial parameters. The defined aim may be at a relatively high level. For instance, in one example the defined aim may be to perform an enzyme titration assay.

[0056] The initial parameters may include any suitable number and range of parameters. For instance, the initial parameters may include DOE (design of experiments) factors and ranges, i.e. defining which variables can be controlled, and to what degree their values can be controlled, in the assay. The initial parameters may include names of a protein, substrate and a reference compound to be used, as well as starting concentrations of each.

[0057] The defined aim and initial parameters may then be input to an assay development module of a computer processor implementing the described method. The assay development module may be part of the scheduling software for controlling equipment to perform assays, or may be separate from the scheduling software. The assay development module is configured to determine various conditions for performing the requested assay based on the defined aim and input parameters.

[0058] It will be understood that values for various conditions for performing an assay may be determined as part of an assay development step / process, and defining different conditions for different aims / assays may be needed. One or more biochemical conditions may be determined. For instance, these may include buffer scouting, substrate / ligand optimal concentration, ideal time point, ideal target concentration, DMSO (dimethyl sulfoxide) concentration, etc. For cell assays, ideal transfection conditions and cell number may be determined. Conditions to validate the assay may also be determined, such as plate uniformity and SAR (structure activity relationship) sets. The conditions to be determined in the assay development module are determined using procedural generation rules, i.e. the conditions are determined in a deterministic manner. In one example, buffer components and ratios for buffer screening, substrate concentration(s) for substrate titration, enzyme concentration(s) and time points for enzyme titration, and maximum DMSO tolerance are determined according to rules-based procedural generation based on the defined inputs. Validation conditions relating to plate uniformity and dose response uniformity may also be defined.

[0059] The determined conditions are used to determine / generate material transfer instructions to be used to automatically prepare the materials for performing the designed assay. As described above, this involves generating reagent and compound plates, under the determined conditions, before selecting and retrieving required samples from automated stores and dispensing material onto the plates according to the material transfer instructions (derived from the determined conditions) using automated liquid handling equipment.

[0060] The assay is run by incubating the materials for the required time and then performing a readout process. The readout may involve different techniques depending on the type of assay being run. Performing the readout may also involve different types of instrumentation, e.g. microscopes, flow cytometer, etc. For instance, this could involve absorbance readouts, fluorescence readouts, etc. The readout is performed by moving the assay plate to a plate reader.

[0061] Various analyses are then performed automatically to determine the performance of the assay. The results of this analysis can be used to update / further develop the assay in order to improve performance, in a deterministic manner as part of a design loop. As an example, during an initial step / run, a signal versus noise score may be determined. This can then be used to determine which DOE factors are taken forward in the assay development process, for instance. In subsequent loops, a surface response analysis (of various variables) may be performed to develop a predictive model of the relationship between the DOE factors and response. Again, this can be used to refine which factors are taken forward, i.e. DOE factors are selected according to deterministic rules using the relationship between DOE factors and surface response. As another example, the output of the assay results analysis can be used in conjunction with Michaelis-Menten kinetics to determine / update the substrate concentrations of the assay being developed. As is known, Michaelis-Menten kinetics describes an enzyme- catalysed reaction rate as a function of substrate concentration. Defined models may be used in this regard to determine updated substrate concentration using reaction rate of a previous run of the assay using defined rules. Further defined models or rules may be used to determine enzyme concentration and time points based on assay analysis results.

[0062] Performing analysis of the assay may include determining values known in the art such as S:B, namely, a signal-to-background noise ratio in the readout, and Z factor, which indicates whether the response in the assay is of sufficient interest to warrant further attention. These values may be used in conjunction with a defined model / rules to determine / update a maximum tolerated DMSO. For instance, a model may interpolate between points to determine a percentage that reduces the signal by less than certain amount, e.g. 10%.

[0063] An assessment of plate uniformity may be performed using the readout from the assay. In this regard, statistics / metrics for the entire plate may be determined, as well as by row and column of the plate. These metrics may include S:B and Z factor, as mentioned above, as well as minimum and maximum edge effect, and minimum and maximum drift effect, in the signal. These metrics may be combined / analysed in a deterministic way to validate the assay, e.g. as a go / no go binary output.

[0064] In a corresponding manner, an assessment of dose response uniformity may be performed using the same metrics as above for plate uniformity and a dose response model fit. This may be performed individually for each loop as well as a single global fit. Again, these may be combined in a deterministic manner, according to defined rules, to validate the assay. A further assessment of dose response may be performed based on a comparison of dose response fitted values for each of a plurality of compounds.

[0065] Updated conditions determined based on the assay readout and results analysis using defined deterministic rules are then used to design / develop an updated assay, which is then run and analysed as above. This can be performed as part of a design loop until one or more stop conditions are satisfied, e.g. the score of one or more metrics, such as the ones outlined above, satisfy certain thresholds, or a prescribed number of development loops have been performed. As mentioned above, the scheduling instructions for performing the assay may also be determined using deterministic rules.

[0066] Figure 2 shows the steps of a method 20 for assaying compounds in examples of the invention. The method 20 comprises computer-implemented steps, in particular implemented on a controller having one or more suitable computer processors, as well as suitable signal inputs and outputs. A system for assaying compounds may include the controller and one or more robotic / automation devices controlled by the controller to implement an assay. The method 20 may comprise, at step 201 , defining a cascade or series of a plurality of assays to be performed. Different assays may be in different tiers / buckets, e.g. tier 1 , tier 2, etc., with the assays becoming more difficult / complex through the tiers. Any suitable number and type of assays may be defined, e.g. enzyme titration assay, binding assay, etc. The assay cascade may be stored in a database accessible by the controller(s) performing the method steps. Information identifying the assays in the database may take various forms, e.g. a natural language description of the assay, an assay ID identifying a specific type of assay, etc.

[0067] At step 202, the method 20 involves initiating the defined assay cascade. This may be prompted by receipt of an indication that a plurality of synthesised compounds, e.g. stored in a suitable compound store, are ready to be assayed according to the defined assay cascade. The synthesised compounds may be received from an automated chemical synthesis platform in a laboratory, for instance. Such an automated chemical synthesis platform may be fully integrated with a system / platform for performing methods of the present invention such that no human intervention, e.g. to move synthesised compounds from one platform to the other, is needed. Step 203 of the method 20 then involves identifying a next assay - initially, a first assay - in the defined assay cascade.

[0068] Step 204 then involves checking whether the identified assay exists or is defined in the database. By this is meant that a standard operating procedure (SOP) for performing the identified assay is stored in the database, where the SOP includes a value for each of a plurality of assay conditions for performing the identified assay. An assay may have any suitable number of defined conditions under which it is to be performed / run, and can include one or more of the conditions outlined above, such as biochemical conditions, cell assay conditions, validation conditions, etc.

[0069] If the identified assay is defined in the database with a defined SOP, then the method 20 proceeds to step 205 to initiate execution of the assay on the received plurality of synthesised compounds. This involves outputting control signals to control various pieces of equipment to perform the defined assay.

[0070] If, however, the identified assay is not defined in the database, then at step 206 the method 20 involves automatically determining an SOP including a value for each of the plurality of assay conditions for performing the identified assay. In particular, this is performed using rules-based procedural generation as described above. The inputs to a procedural generation module of the controller / system include the reagent names and concentrations, e.g. defined by a user. The procedural generation module determines candidate values for each of a plurality of conditions under which the assay is to be performed.

[0071] An assay development step is then performed in which the assay is run with the candidate values for each of the conditions. This may be performed using suitable compounds, e.g. some of the received plurality of compounds. The readout from the assay is then analysed - automatically, semi-automatically, or manually - to determine a plurality of performance metrics to determine performance of the assay with the candidate condition values. If the determined metric values / outputs satisfy respective thresholds / requirements - indicating acceptable performance of the developed assay - then the assay can be finalised. If needed, the finalised assay can be run using the received plurality of compounds at step 205.

[0072] Returning to step 205 of the method 20, the controller may execute the defined assay by initiating defined scheduling instructions for controlling the equipment needed to run the assay based on the defined SOP. In particular, this may involve controlling one or more robotic devices to perform various tasks. These devices could include robotic arms and conveyors for moving samples, for instance. The scheduling instructions for controlling various robotic equipment may take any suitable form. Examples may include, but are not limited to: gathering reagents from one or more stores (that store reagents at different temperatures, e.g. -80, -20, 4 degrees Celsius); taking reagents to a pipetting station; gathering plasticware from a store (e.g. an ambient temperature store); pumping liquids onto pipetting deck; preparing assay buffer using reagents, liquids and plasticware; preparing assay reagents at ideal concentrations from assay buffer; pipetting assay reagents into a dispensing plate; moving the dispensing plate to the dispenser; trashing all of the remaining reagents / plasticware; dispensing assay reagents into an assay plate; incubating the assay plate and adding more reagents, as needed; moving the assay plate to a plate reader; trashing the dispensing plate; and reading the assay in the plate reader. Gathering the reagents from a store typically involves collecting tubes from an automated tube store and puts / deposits them is a defined / special plate. Preparing the assay buffer typically involves using an automated pipette and moving a certain volume of liquid to a certain position. Dispensing the assay reagents typically involves calling a specific dispensing protocol and includes setting an ideal pressure for the liquid class under consideration.

[0073] With continuing reference to step 205, if defined scheduling instructions for implementing the defined SOP to run the defined assay, then the method 20 may involve automatically determining scheduling instructions for instructing the robotic devices to perform the process steps of the defined SOP. In particular, the scheduling instructions for performing the process steps may be determined using rules-based procedural generation based on the defined SOP of the assay to be run.

[0074] The results of the assay may be analysed to determine one or more properties of the synthesised compounds. This may then be used to inform which of the compounds to take forward to the next assay in the assay cascade. This selection may be performed automatically or manually.

[0075] The method 20 then returns to step 203 to identify the next assay in the cascade. The identified next assay is to be used to assay the compounds selected to be taken forward based on the results of the previous assay in the assay cascade. The remaining steps 204, 205, 206 are then performed, as needed, in order to perform this next assay in the assay cascade. In particular, in a corresponding manner to above, the SOP for performing the assay is either retrieved from memory or determined automatically using rules-based procedural generation. Also, the scheduling instructions for initiating the process steps to be performed by various equipment to perform the assay are either retrieved from memory or determined automatically, based on the defined SOP, using rules-based procedural automation. The repeated steps of the method 20 are repeated until all of the assays in the assay cascade have been performed. The assay cascade may therefore be run / implemented with minimal or zero human input / intervention.

[0076] Hardware for performing the described methods may be in the form of any suitable computing device, for instance one or more functional units or modules implemented on one or more computer processors. Such functional units may be provided by suitable software running on any suitable computing substrate using conventional or custom processors and memory. The one or more functional units may use a common computing substrate (for example, they may run on the same server) or separate substrates, or one or both may themselves be distributed between multiple computing devices. A computer memory may store instructions for performing the methods performed by the controller, and the processor(s) may execute the stored instructions to perform the methods.

[0077] Illustrative examples are provided as follows. In an enzymatic assay, for instance, in a oneway titration step a Km parameter (i.e. substrate concentration at half the maximum reaction velocity) may need to be determined. This may be determined with reference to a defined R2cut off, e.g. between 0.3 and 1 , where R2is a coefficient of determination. The Km providing the best R2may be selected.

[0078] Figure 3 schematically illustrates procedural generation steps for developing a substrate titration step / process. It is determined whether it is a one-way or two-way substrate titration. For one-way, a first model, ‘Model T, is used and a fit of Model 1 with the uploaded data is performed with optional removal of outliers. For two-way titration, Model 1 or a second model, ‘Model 2’, may be used to fit the uploaded data with optional removal of outliers. The final Kms (for two-way titration only) are determined. Substrate plate quality control is then performed, e.g. a pass if R2is between 0.3 and 1 . An example definition of Model 1 , along with example initial values, is shown in Figure 4. An example definition of Model 1 , along with example initial values, is shown in Figure 5.

[0079] Table 1

[0080] Table 1 shows an example of substrate titration automated model selection. In particular, a comparison of R2for a Y=0 model and a Michaelis Menten model is performed on a 0 concentration trace for both substrates, with the model selection based on the R2value being indicated in the table. R2may be calculated as 1 -RSS / TSS, where RSS is residual sum of squares and TSS is total sum of squares. The Michaelis Menten model may be defined as shown in Figure 6.

[0081] For the model fit in Figure 3, a standard model used may be least squares regression. Both Models 1 and 2 may be fit in logarithmic space. Baseline subtraction may be performed by subtracting a last column (control) from all columns. In terms of model choice, for one-way titration Model 1 may always be chosen. For two-way titration the choice may be according to: Model 1 when there is a background reaction when S1 >0 and S2>0; Model 2 when there is no background reaction when S1 >0 and S2>0.

[0082] For automated outlier detection, a ROUT method may be used, as described here: https: / / bmcbioinformatics.biomedcentral.eom / articles / 10.1186 / 1471 -2105-7-123. This involves a robust least squares regression using cauchy loss. The FDR (False Discovery Rate) method is applied as described in the provided reference to detect outliers, optionally using the following parameter values: Q=1 %; K=3 for one-way titration model 1 ; K=7 for two-way titration model 1 ; K=5 for two-way titration model 2. Use correction for RSDR (Robust Standard Deviation of the Residuals) for when N is small when N<15. Detected outliers may be displayed as x’s. If user is happy then rerun fit to run standard least squares regression or can exclude more points / reinclude points then rerun. Final KMs (for two- way titration only) may be the geometric mean of kms: Final km1 =(col km1 * row km2)2; Final km2=(row km1 * col km2)2.

[0083] In a binding assay, for instance, in a probe titration and time course step a Kd parameter (i.e. equilibrium disassociation constant) and incubation time may need to be determined. Figure 7 schematically illustrates procedural generation steps for developing a probe titration and time course step. Example model definition parameters and optional initial values are provided in Figure 8. The model fit may be least squares regression. The model may be fit in logarithmic space. Baseline subtraction may be performed by subtracting a last column (control) from all columns

[0084] There may need to be less than a defined difference between Kd between time points, e.g. less than a two-fold difference. A Bmax parameter (i.e. maximum observed binding) may also need to be less than a defined level, e.g. less than 1 .2 fold. The determination may involve a consideration of cases in which Bmaxis less than the defined level / value. In particular, Kd may be determined for the curves in each of these cases. The lowest Kd may then be selected as the Kd parameter (provided it is less than the defined difference, e.g. less than a two-fold difference). Plate quality control (QC) may be achieved if R2is between 0.3 and 1 for all models.

[0085] In an antibody / probe versus protein titration step an antibody / probe and protein concentration may need to be determined. Figure 9 schematically illustrates procedural generation steps for developing an antibody / probe versus protein titration step. For instance, the recommended / determined protein concentration may be the lowest concentration that gives an S:B value greater than a defined value, e.g. greater than or equal to three. Within that constraint for protein concentration, the recommended / determined antibody concentration may be the lowest concentration that gives an S:B value greater than a defined value, e.g. greater than or equal to three. In an unlabelled probe inhibition step, a pass may be determined if an IC50 value is less than a defined value, which may be a function of the Kd parameter, e.g. less than or equal to 3*Kd. For direct binding assays there is no antibody and so it is either a protein titration only with optional probe titration.

[0086] In some examples, recommended DMSO tolerance may be set to be a defined value, e.g. 1 %, unless the observed value is less than a certain value, e.g. 0.5% (in which case human intervention may be needed). Figure 10 schematically illustrates procedural generation steps for developing a recommended DMSO tolerance. The process may involve data preparation, involving taking an average of multiple (e.g. four) repetitions, dividing against 0% DMSO to obtain signal-to-background (bg) window. The recommended maximum DMSO percentage is the maximum percentage that reduces the signal by less than 10%. Hence, DMSO percentages are interpolated to find the percentage at 90% of the maximum signal.

[0087] Plate statistics in the form of plate uniformity, dose response (DR) uniformity and dose response validation, as well as unlabelled dose response, may be determined. DR model fit may be the same as the core plate runner. For unlabelled probe DR, there may be comparison between two runs of the same plate. Plate QC may be as follows. An unlabelled probe DR checkpoint may be that IC50 should be 3-fold of Kd. For DR validation and uniformity, S:B, Z, CV% (coefficient of variation), edge effect and drift effect may need to satisfy certain thresholds checkpoints for a pass. In one purely illustrative example, S:B> 2.5, Z>0.6, CV%<10, xC50 between repetitions is within 3-fold of the geometric mean, and percentage edge and drift effect is less than or equal to ±10%. For a second plate, a slope for correlation is greater than 0.8 and IC50 is within 3-fold between the two plates. Similar checkpoints may be set for plate uniformity. A plate may pass or fail based on statistics / metrics for the entire plate. Many modifications may be made to the described examples without departing from the scope of the appended claims.

[0088] In the examples described above, an assay cascade is implemented whereby each assay is performed in sequence and, if an SOP for a specific assay is not stored / available, then the SOP is determined automatically using procedural generation when the cascade reaches said specific assay in the sequence. It will be understood, however, that in different examples it may be determined prior to implementing any of the assays in an assay cascade whether an SOP is stored / available for each assay in the cascade and, if not, then determining the SOP for each relevant assay automatically using procedural generation prior to initiating the assay cascade.

Claims

CLAIMS1. A computer-implemented method of controlling a system for assaying compounds, the method comprising: defining an assay cascade comprising a plurality of assays to be performed in a defined sequence; receiving an indication that a plurality of synthesised compounds are to be assayed according to the defined assay cascade; identifying a first assay, in the defined sequence, of the assay cascade; accessing an assay database storing a plurality of defined assays each including a standard operating procedure, SOP, for performing the respective defined assay, wherein each SOP includes a value for each of a plurality of assay conditions for performing the respective defined assay; determining whether the first assay is one of the plurality of defined assays stored in the assay database; if the first assay is one of the plurality of defined assays, then retrieving the SOP including the values of the plurality of assay conditions of said one of the defined assays from the assay database; if the first assay is not one of the plurality of defined assays, then determining an SOP including a value for each of the plurality of assay conditions for performing the first assay; controlling the system to perform the first assay on the plurality of synthesised compounds according to the SOP of the first assay, and obtaining one or more results of the first assay from the system, wherein determining the SOP for performing the first assay comprises: retrieving a plurality of defined SOP procedural rules for determining SOPs; and processing the plurality of defined SOP procedural rules to generate the SOP for performing the first assay, the method further comprising repeating steps of: identifying a next assay, in the defined sequence, of the assay cascade; determining a set of compounds, from the plurality of synthesised compounds, on which the next assay is to be performed, wherein the set of compounds is determined based on the one or more results of the previously-performed assay obtained from the system;accessing the assay database, and determining whether the next assay is one of the plurality of defined assays stored in the assay database; if the next assay is one of the plurality of defined assays, then retrieving the SOP including the values of the plurality of assay conditions of said one of the defined assays from the assay database; if the next assay is not one of the plurality of defined assays, then determining an SOP including a value for each of the plurality of assay conditions for performing the next assay, wherein determining the SOP for performing the next assay comprises retrieving the plurality of defined SOP procedural rules for determining SOPs and processing the plurality of defined SOP procedural rules to generate the SOP for performing the next assay; controlling the system to perform the next assay on the set of compounds according to the SOP of the next assay, and obtaining one or more results of the next assay from the system, until each of the assays in the assay cascade has been performed.

2. A method according to Claim 1 , wherein the plurality of defined SOP procedural rules are processed based on one or more defined initial parameters, including: one or more variables that can be adjusted and a range of values within which each respective variable can be adjusted.

3. A method according to Claim 2, wherein the variables that can be adjusted include one or more of: a concentration of a protein to be used in the assay; a concentration of a substrate to be used in the assay; and a concentration of a reference compound to be used in the assay.

4. A method according to any previous claim, wherein the plurality of assays include one or more of: an enzyme titration assay; and a binding assay.

5. A method according to any previous claim, wherein the plurality of assay conditions includes one or more biochemical conditions; optionally, wherein the biochemical conditions include one or more of: buffer scouting; substrate optimal concentration; ideal time point; ideal target concentration; and dimethyl sulfoxide concentration.

6. A method according to any previous claim, wherein the plurality of assay conditions includes one or more cell assay conditions; optionally, wherein the cell assay conditions include one or more of: ideal transfection conditions; and cell number.

7. A method according to any previous claim, wherein the plurality of assay conditions includes one or more validation conditions for validating assay performance; optionally, wherein the validation conditions include one or more of: plate uniformity; and structure activity relationship sets.

8. A method according to any previous claim, wherein processing the plurality of defined SOP procedural rules to generate the SOP for performing the first assay is performed as part of an iterative process, comprising: processing the plurality of defined SOP procedural rules to generate a candidate SOP including a candidate value for each of the plurality of assay conditions for performing the first assay; controlling the system to perform the first assay on one or more reference compounds according to the candidate SOP of the first assay, and obtaining one or more intermediate results of the first assay from the system; determining a value for each of one or more parameters each indicative of effectiveness of a respective aspect of the first assay, wherein the parameter values are determined based on the obtained intermediate results; and determining whether each determined parameter value satisfies a respective desired parameter condition, wherein if the determined parameter values satisfy the respective desired parameter conditions, then the candidate SOP is determined to be the SOP for performing the first assay, wherein if one or more of the determined parameter values do not satisfy the respective desired parameter conditions, then the method comprises repeating steps of: processing the plurality of defined SOP procedural rules, based on the parameter values obtained from performing the first assay according to the previous candidate SOP, to generate an updated candidate SOP including an updated candidate value for each of the plurality of assay conditions for performing the first assay; controlling the system to perform the first assay on one or more reference compounds according to the updated candidate SOP of the first assay, and obtaining one or more intermediate results of the first assay from the system; anddetermining, based on the obtained intermediate results, the parameter values indicative of effectiveness of respective aspects of the first assay, until a stop condition is satisfied, wherein the updated candidate SOP when the stop condition is satisfied is determined to be the SOP for performing the first assay.

9. A method according to Claim 8, wherein the reference compounds include the plurality of compounds on which the first assay is to be performed.

10. A method according to Claim 8 or Claim 9, wherein the stop condition is one of: the repeated steps have been performed a defined number of times; and a rate of improvement of the determined parameter values falls below a defined threshold rate of improvement.

11. A method according to any of Claims 8 to 10, wherein the parameters indicative of assay effectiveness include one or more of: a signal-to-noise ratio parameter; a surface response parameter; a Michaelis-Menten kinetics parameter; a Z factor parameter; a plate uniformity parameter; and a dose response uniformity parameter.

12. A method according to any previous claim, wherein controlling the system to perform the first assay on the plurality of synthesised compounds according to the SOP of the first assay comprises: retrieving a plurality of defined system steps procedural rules for determining process steps to be implemented by the system to perform assays; processing the plurality of defined system steps procedural rules, based on the SOP of the first assay, to generate the process steps for performing the first assay; and instructing the system to implement the generated process steps to perform the first assay.

13. A method according to Claim 12, wherein controlling the system to perform the next assay on the determined set of compounds according to the SOP of the next assay comprises: retrieving the plurality of defined system steps procedural rules, and processing the plurality of defined system steps procedural rules, based on the SOP of the next assay, to generate the process steps for performing the next assay; andinstructing the system to implement the generated process steps to perform the next assay.

14. A method according to any previous claim, wherein controlling the system to perform the first or next assay comprises controlling one or more robotic devices of the system to retrieve and prepare materials needed to perform the first or next assay.

15. A non-transitory, computer readable storage medium storing instructions thereon that, when executed by one or more computer processors, cause the one or more computer processors to perform a method according to any previous claim.

16. A controller comprising one or more computer processors, the controller being configured to perform a method according to any of Claim 1 to 14.

17. A system for assaying compounds, the system comprising: a controller according to Claim 16; and one or more robotic devices configured to receive scheduling instructions from the controller to perform one or more actions to implement each assay in the assay cascade.

Images