Method for testing active ingredients

Abstract

The invention refers to a computer implemented method for testing active ingredients with respect to an interaction strength with a target protein structure. For candidate active ingredients a digital twin is received. A digital twin associated with a molecular cut-out of a target protein structure is received. Conformations associated with the candidate active ingredients are determined based on the digital twin associated with the respective candidate active ingredients. An energy measure for the candidate active ingredients is determined based on the conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the interaction site with the target protein structure. Based on the determined energy measure, digital twins associated with one or more of the candidate active ingredients for further testing with respect to one or more application goals.

Description

[0001] Method for testing active ingredients

[0002] FIELD OF THE INVENTION

[0003] The invention refers to a method, apparatus and computer program product for testing active ingredients with respect to an interaction strength with a target protein structure. Moreover, the invention refers to a test method, apparatus and computer program product for testing active ingredients with respect to one or more application goals.

[0004] BACKGROUND OF THE INVENTION

[0005] For finding new or improving known active ingredients, for instance, for agricultural or medical applications, a plurality of candidate active ingredients have to be tested in long and resource-intensive test series. Thus, improving and accelerating the testing procedure would be advantageous and could lead to a faster development of new and improved active ingredients.

[0006] SUMMARY OF THE INVENTION

[0007] The invention allows for an improved testing procedure for testing candidate active ingredients with respect to one or more predetermined application goals, for instance, agricultural or medical applications.

[0008] In a first aspect, a computer implemented method is presented fortesting active ingredients with respect to an interaction strength with a target protein structure, wherein the method comprises a) receiving for one or more candidate active ingredients a digital twin associated with a respective candidate active ingredient, b) receiving a digital twin associated with a molecular cut-out of a target protein structure, wherein the molecular cut- out comprises an interaction site between the target protein structure and an active ingredient, c) determine one or more conformations associated with the one or more candidate active ingredients based on the digital twin associated with the respective one or more candidate active ingredients, d) determining an energy measure for the one or more candidate active ingredients based on the one or more conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the interaction site with the target protein structure, e) providing the energy measure for selecting one or more of the candidate active ingredients for further testing with respect to one or more application goals.

[0009] In a further aspect, a computer implemented method is presented for testing active ingredients with respect to an interaction strength with a target protein structure, wherein the method comprises a) receiving for one or more candidate active ingredients a digital twin associated with a respective candidate active ingredient, b) receiving a digital twin associated with a molecular cut-out of a target protein structure, wherein the molecular cutout comprises an interaction site between the target protein structure and an active ingredient, c) determine one or more conformations associated with the one or more candidate active ingredients based on the digital twin associated with the respective one or more candidate active ingredients, d) determining an energy measure for the one or more candidate active ingredients based on the one or more conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the interaction site with the target protein structure, e) selecting, based on the determined energy measure, one or more of the candidate active ingredients for further testing with respect to one or more application goals, f) generating production data for controlling and / or monitoring a production of the selected one or more of the candidate active ingredients for further testing.

[0010] In a further aspect, a test method is presented for testing active ingredients with respect to one or more application goals, wherein the method comprises a) performing a method as described above for determining one or more active ingredients with a predetermined interaction strength with a target protein structure, b) producing the one or more active ingredients based on the provided digital twins of the one or more active ingredients, c) subjecting the produced one or more active ingredients to further testing to determine an applicability of the one or more active ingredients with respect to one or more application goals. In a further aspect, an apparatus is presented for testing active ingredients with respect to an interaction strength with a target protein structure, wherein the apparatus comprises one or more processors configured to perform a method comprising a) receiving a digital twin associated with one or more candidate active ingredients, b) receiving a digital twin associated with a molecular cut-out of a target protein structure, wherein the molecular cutout comprises an interaction site between the target protein structure and an active ingredient, c) determine one or more conformations associated with one or more candidate active ingredients based on the digital twin associated with the respective one or more candidate active ingredients, d) determining an energy measure for the one or more candidate active ingredients based on the one or more conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the binding site with the target protein structure, e) providing, based on the determined energy measure, digital twins for one or more active ingredients of the candidate active ingredients for further testing with respect to one or more application goals.

[0011] In a further aspect, a test system is presented for testing active ingredients with respect to one or more application goals, wherein the system comprises a) an apparatus as described above for performing a method as described above for determining one or more active ingredients with a predetermined interaction strength with a target protein structure, b) a production facility configured for producing the one or more active ingredients based on the provided digital twins of the one or more active ingredients, c) a testing facility configured for subjecting the produced one or more active ingredients to further testing to determine an applicability of the one or more active ingredients with respect to one or more application goals.

[0012] In a further aspect, a computer program product is presented fortesting active ingredients, wherein the computer program product causes an apparatus as described above to carry out the method as described above when executed by the apparatus.

[0013] In a further aspect, a computer program product is presented fortesting active ingredients, wherein the computer program product causes a system as described above to carry out the method as described above when executed by the system.

[0014] Since one or more conformations associated with the one or more candidate active ingredients are determined based on the digital twin associated with the respective one or more candidate active ingredient and since the energy measure is determined based on the one or more conformations, the active ingredients with the highest likelihood of interaction, e.g. binding, to the interaction site of the target protein structure can be determined more reliably. In particular, taking automatically the most likely conformations of the active ingredient into account allows to determine the energy measure for the most likely interaction between the active ingredient and the target protein. Accordingly, the determined energy measure of the interaction strength leads to more reliable results when testing the active ingredients with the highest likelihood of interacting in a desired manner with the target protein. Moreover, since the interaction strength correlates with the effect of the active ingredient on the protein, providing active ingredients with the respective interaction strength more reliably avoids to utilize testing resources on candidate active ingredients that do not interact in a desired predetermined manner to the protein and thus cannot fulfil a respective technical application goal. Hence, also the overall testing procedure for finding new or improved active ingredients can generally be improved, in particular, accelerated, made more reliable and less resource-intensive.

[0015] An active ingredient can be any molecule that provides an intended target effect on a target protein. The target effect can be coupled to a technical application of the active ingredient, for instance, in the context of agricultural or medical applications. The active ingredient can have an effect on proteins of biological systems that are of human, animal, fungi, or plant origin. The active ingredient can be a small molecule. The target protein structure is associated with a protein that has been identified as target for reaching a technical goal in a system comprising proteins. For example, the system can be of animal, fungi, human or plant origin and the technical effect can correspond to increasing or decreasing an activity of the protein in the system. However, the system can also be a waste water environment comprising protein and the technical effect can correspond to destroying or deactivating the protein. An interaction strength between the active ingredient and the target protein structure refers to the strength of the interaction between the active ingredient and the target protein structure. The interaction strength can be represented by an interaction energy indicative of the strength of the interaction between the active ingredient and the target protein structure. The interaction strength can refer to a binding strength or associated strength and the interaction energy to a binding energy or an associated energy.

[0016] The method comprises receiving for one or more candidate active ingredients a digital twin associated with the respective one or more candidate active ingredients and also receiving a digital twin associated with a molecular cut-out. The receiving can refer to any computer process that provides the respective digital twin for further processing in accordance with the defined method. For example, the receiving can refer to accessing a storage on which the respective digital twin is stored and providing the digital twin for further processing. The receiving can also refer to receiving the respective digital twin via other digital communication pathways, for instance, from a user interface in which a user indicates the respective digital twin.

[0017] A digital twin represents a real-world object in the digital domain. The digital twin can represent all or only predetermined aspects of the respective real-world object. The digital twin associated with the one or more candidate active ingredients represents one or more aspects of the candidate active ingredients. For each candidate active ingredient one digital twin can be received. The digital twin can represent a molecular structure of the respective associated candidate active ingredient. For example, the digital twin can include a SMILES representation, an International Chemical Identifier, an MDL molefile, an embedding model representation, or a Cartesian model representation of the active ingredient. In an embodiment, the digital twin associated with the one or more active ingredients can include a Cartesian model that defines Cartesian coordinates and element type for the atoms of a respective candidate active ingredient.

[0018] The digital twin of the molecular cut-out can also include a MDL molefile, e.g. a PDB file, or a Cartesian model representation of the molecular cut-out. In an embodiment, the digital twin associated with the molecular cut-out includes a Cartesian model representation that defines for the atoms of the molecular cut-out Cartesian coordinates of their positions and the element type. The molecular cut-out of the target protein structure refers to a part of the target protein structure that has been cut-out from the target protein structure by virtually cutting through the molecular bonds between the molecular cut-out and the rest of the target protein structure. The molecular cut-out hence focuses on a specific part of the target protein structure that comprises the interaction site between the target protein structure and the active ingredient. The interaction site between the target protein structure and the active ingredient can refer to any site that has been identified as a site on the target protein structure that interacts with an active ingredient. For example, the interaction site can include a binding pocket. However, the interaction site can also be situated on the outside of the target protein structure. The molecular cut-out together with a bound candidate active ingredient can define a molecular cluster.

[0019] In an embodiment, the molecular cut-out is generated by cutting the target protein molecular structure in a predetermined radius around to the interaction site. The radius can be predetermined based on literature values, experiments or experience of the respective operator. For example, for most proteins and application cases a predetermined radius between 5 and 6 Angstrom can be suitable. The method further comprises to determine one or more conformations associated with the one or more candidate active ingredients. In particular, the one or more conformations are automatically determined. For each of the one or more candidate active ingredients one or more conformations can be determined. Conformations are molecular structures that are caused by conformational isomerism, wherein the conformations of the molecule can be interconverted just be rotations around formally single bonds. The determined one or more conformations can include conformations of the active ingredients that correspond to one or more conformations associated with a local minimum on a potential energy surface of the active ingredient. The conformations of a candidate active ingredient can also include one or more conformations referring to a stable conformation that can be isolated. The one or more conformations of a candidate active ingredient are determined based on the respective digital twin associated with the candidate active ingredient. The conformations can be determined utilizing, for instance, molecular dynamic methods or angle-dependent methods. In an embodiment, the conformations of a candidate active ingredient are determined based on the digital twin associated with the candidate active ingredient by utilizing a molecular dynamic method.

[0020] In an embodiment, the determining of one or more conformations comprises, selecting one or more conformations based on an energy of the conformations associated with a candidate active ingredient. The energy can be a free energy that is indicative of the potential energy of the respective conformation. For example, the free energy can refer to the Gibbs free energy or the Helmholtz free energy. The conformations can be selected based on the free energy by comparing a free energy of a conformation to a predetermined free energy threshold and based on the comparison selecting the associated conformation. For example, conformations with a free energy below a predetermined free energy threshold can be selected. Moreover, conformations associated with a minimum of the free energy can be selected. Further, the energy can be an electronic energy of the conformations indicative of the electronic energy level of the conformations. The conformation can be selected based on the electronic energy level by choosing one or more of the conformations with the lowest electronic energy level compared to the other conformations or by choosing conformations with an electronic energy level below a predetermined threshold.

[0021] The energy measure can be determined for each of the one or more candidate active ingredients including determining an energy measure for each of the one or more conformations of each of the one or more candidate active ingredients. Moreover, determining an energy measure for a candidate active ingredient can comprise determining an energy measure for the respective one or more conformations. An energy measure can then be determined for the candidate active ingredient by combining the energy measures of the one or more conformations. For example, a weighted average can be utilized for determining the energy measure of the active ingredient, wherein the weighted average weights the energy measures determined for respective one or more conformations by a predetermined weight associated with the respective one or more conformations. For example, the weight can be associated with a likelihood of existing a predetermine time period, an average existing time, and / or a fraction of a conformation on the population of an active ingredient. In case one or more conformations are selected, the same embodiment can be applied. Alternatively, in case of selected one or more conformations the energy measure of the selected conformers can also refer to the energy measure of the active ingredient, or an energy measure or a conformer can be selected as energy measure of the active ingredient. The energy measure can be selected, for example, by selecting the lowest or highest energy measure of energy measures of the conformations, or by selecting the energy measure of the most likely conformation. The energy measure can be any measure that is indicative of the interaction strength between an active ingredient interacting at the interaction site with the target protein structure. For example, the energy measure can include the Gibbs free energy. Further, the energy measure can include at least one of an electronic binding energy, a single-point gas phase energy, a translational, rotational and / or vibrational enthalpy, a translational, rotational and / or vibrational entropy, and a solvation energy. For determining the energy measure, for instance, semi-empiric methods, machine learning potentials, or density functional theory methods can be utilized. Preferably, the energy measure is a relative energy measure, more preferably a relative binding energy. A relative energy measure for a candidate active ingredient can be defined as the difference between an absolute energy measure for the candidate active ingredient and an absolute energy measure for a predetermined reference. The reference can be any molecule that binds to the interaction site of the target molecular structure. For example, the candidate active ingredient and the predetermined reference can be both ligands of the target protein structure. Generally, the reference can be chosen arbitrarily or based on literature references, experience or other pre-knowledge.

[0022] The energy measure is then provided for selecting one or more candidate active ingredients for further testing. The testing includes producing the one or more candidate active ingredients. Preferably, digital twins of one or more of the candidate active ingredients are selected based on the energy measure and are provided for further testing with respect to one or more application goals. The provided digital twins associated with the one or more candidate active ingredients can include machine executable instructions for producing the respective one or more candidate active ingredients. The providing of the digital twins associated with the one or more candidate active ingredients based on the determined energy measure can comprise selecting one or more of the candidate active ingredients as test active ingredients based on the determined energy measure and providing the digital twins associated with the test active ingredients for further testing including producing the test active ingredients. Further the provided digital twins associated with the test active ingredients can be configured for further testing including producing the test active ingredients, for example, can include respective production instructions and optionally testing instructions. The selecting can be based on the determined energy measure and a target energy measure. For example, based on a comparison between the determined energy measure and a target energy measure. A candidate active ingredient can be selected as test active ingredient if the associated energy measure meets the respective target energy measure. For example, the target energy measure can be an energy measure range that defines the interaction strength determined as being most suitable for the intended technical goal of the active ingredient. The providing can further include providing production data comprising instructions with respect to the production of one or more of the digital twins associated with the one or more candidate active ingredients that are selected. The production data can be configured for controlling and / or monitoring a production ofthe respective selected candidate active ingredients. The production data can refer to any data that allows to derive respective controlling and / or monitoring signals from the production data. The such produced candidate active ingredients can then be subjected to further testing with respect to the one or more predetermined application goals of the active ingredient.

[0023] In an embodiment, the method further comprises determining one or more conformations associated with one or more active ingredients interacting at the interaction site with the molecular cut-out based on the digital twin associated with respective one or more candidate active ingredients and the digital twin associated with the molecular cut-out, wherein the energy measure is further determined based on the one or more conformations associated with one or more active ingredients interacting at the interaction site with the molecular cut-out. This embodiment allows not only to determine conformations for an active ingredient alone but to determine conformations for the active ingredient interacting with the target protein structure at the interaction site. Thus, conformations that only occur when the active ingredient is interacting with the target protein structure can be taken into account for determining the energy measure. Preferably, one or more conformations are selected based on a free energy from the conformations determined for the active ingredient not-interacting with the molecular cut-out and the determined conformations of active ingredients interacting at the interaction site with the molecular cut-out. This allows to select the conformations that are most strongly and hence most likely to interact with the molecular cut-out. This increases the reliability of the results during the further testing of the active ingredients.

[0024] In an embodiment, the energy measure for a conformation of an active ingredient is determined based on an energy difference between an energy of the conformation and the molecular cut-out without interaction and an energy of a respective target protein cluster, wherein the target protein cluster is associated with a conformation interacting with the interaction site of the molecular cut-out. An energy of the conformation and the molecular cut-out without interaction can be determined, for instance, based on determining an energy of the conformation and determining an energy of the molecular cut-out and combining both. Thus, the determined energy measure is indicative of an energy difference between an interacting state of the active ingredient with the respective target molecular structure and a non-interacting state of the active ingredient with the target molecular structure. Such an energy difference is regarded as indicating the interaction strength between an active ingredient and a target molecular structure. The determined energy in this embodiment can be at least one of an electronic binding energy, a single-point gas phase energy, a translation, rotation and / or vibrational enthalpy, a translational, rotational and / or vibrational entropy and a solvation energy. Preferably, the energy difference is determined for at least two of the above energies and the energy measure is determined based on the at least two determined energy differences. Even more preferably, the energy difference is determined as a difference of the single-point gas phase energy, a translational, rotational and / or vibrational entropy and / or enthalpy, and a solvation energy and the energy measure is determined based on the respective determined energy differences. The energy difference can also be a free energy, preferably, a Gibbs free energy.

[0025] In an embodiment, the energy measure for a conformation of an active ingredient is determined based on at least one of an electronic binding energy, a single-point gas phase energy; a translational, rotational, and / or vibrational enthalpy; a translational, rotational, and / or vibrational entropy; and a solvation energy. Preferably, the energy measure is determined based on at least two of the above-mentioned energies. In an embodiment, based on at least two of the above energies a free energy, preferably a Gibbs free energy, is determined as energy measure. For all the above cases the energy measure is preferably determined as relative energy measure, for example, as relative Gibbs free energy.

[0026] In an embodiment, the energy measure is determined based on at least two of a) a electronic binding energy, b) a translational, rotational, and / or vibrational enthalpy and / or a translational, rotational, and / or vibrational entropy, and c) a solvation energy. Preferably, for determining the respective at least two energy measure contributions different methods are utilized, for example the utilized methods can refer to different levels of theory. In quantum chemistry a level of theory of a method refers to the accuracy with which a respective eigenvalue problem can be solved for a specific basis set utilized by the method. For example, for determining translational, rotational, and / or vibrational enthalpy and / or a translational, rotational, and / or vibrational entropy energy contributions a GFN-xTB method, for a solvation energy contribution a COSMO method, and for the electronic binding energy a DFT method can be utilized, wherein all of these methods have a differ in their underlying level of theory.

[0027] In an embodiment, the method further comprises optimizing a geometry of the molecular cut-out and / or a target protein cluster, wherein the target protein cluster is associated with a conformation interacting with the binding site of the molecular cut-out, wherein the energy measure is determined based on the digital twin of the respective optimized molecular cutout and / or the target protein cluster. Due to the molecular cut-out that does not involve the complete target protein structure, the geometry, for instance, represented by the positions of the atoms of the molecular cut-out, can be in an energetically not optimal configuration. Optimizing the geometry allows to determine the energy measure based on an energetically optimal configuration of the molecular cut-out. The geometry optimization, for instance, comprises amending the positions of the atoms of the respective molecular cutout. The energy optimization can be based on determining the geometric structure of the molecular cut-out with the lowest energy. Moreover, the energy optimization can include minimizing the energy with respect to nuclear displacements. For example, the energy function E=f(x,y,z) can be minimized such that dE / dxdydz is smaller than a predetermined threshold near zero, for instance, 10-5Hartree / Bohr.The starting point of the optimization can be a crystal structure of the molecular cut-out and the optimization can be based on relaxing the positions of the atoms in the crystal structure. Preferably, for optimizing of the geometry of the molecular cut-out and / or the target protein cluster the positions of atoms belonging to at least one atom class are constraint. Preferably, the constrained at least one atom class defines the structure of the interaction site of the target protein cluster. For example, C-atoms define the structure more strongly than H-atoms. The constrained at least one atom class can referto any of C-alpha atoms, C-atoms, N-atoms, or combinations thereof. Moreover, all atoms except hydrogen atoms can be constrained. Constraining at least one atom class, preferably, comprising C-atoms, of the molecular cut-out allows, during the geometric optimization, to keep the interaction site structured in accordance with the structure of the interaction site of the complete protein structure. In particular, in case the interaction site refers to an interaction pocket, a change or collapse of the pocket can be prevented by this constraint. In an embodiment, the method further comprises optimizing a geometry of the conformations of the one or more active ingredients, wherein the energy measure is determined based on the digital twin of the respective optimized conformations. The same methods utilized for geometrical optimization of the molecular cut-out can be utilized for the conformations of the active ingredients.

[0028] In an embodiment, the receiving of the digital twin of the molecular cut-out, comprises completing the molecular cut-out by adding predetermined atoms or atom groups to bonds that have been cut. The respective completing of the molecular cut-out by adding predetermined atoms to the bonds that have been cut can be performed by utilizing respective rules for adding respective atoms or atom groups. For example, the rules can take into account the atom of the molecular cut-out to which the atom or atom group should be added, the remaining bonds of this atom, or an environment of this atom. For example, a rule can determined that covalent bonds that have been cut are saturated with a methylated N- or C- terminus of the protein, thereby restoring parts of the proteins backbone.

[0029] It shall be understood that the methods as described above, the apparatuses as described above and the computer program products as described above have similar and / or identical preferred embodiments, in particular, as defined in the dependent claims.

[0030] It shall be understood that a preferred embodiment of the present invention can also be any combination of the dependent claims or above embodiments with the respective independent claim.

[0031] These and other aspects of the present invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

[0032] BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Fig. 1 shows schematically and exemplarily a screening method for screening for a new drug,

[0034] Fig. 2 shows schematically and exemplarily a method for determining and producing a target active ingredient with respect to a predetermined technical application of the active ingredient in a system comprising protein, Fig. 3 shows schematically and exemplarily a visualization of a protein drug complex,

[0035] Figs. 4 and 5 show schematically and exemplarily different aspects of a method for determining an interaction strength of an active ingredient with an interaction site of a target protein,

[0036] Fig. 6 shows schematically and exemplarily a data structure of data utilized in the method,

[0037] Figs. 7 and 8 show examples of free energies determined with the screening method compared to literature values, and

[0038] Figs. 9a to 9e show further examples of free energies determined with the screening method compared to literature values.

[0039] DETAILED DESCRIPTION OF EMBODIMENTS

[0040] In the following examples are provided for technical applications, in which the invention can improve the testing and design process of new or amended active ingredients.

[0041] Protease is an integral part of modern laundry detergent to ensure efficient removal of protein based stains such as grass, blood and egg. In order to ensure efficacy of the protease after storage as well as to enhance stability of secondary enzymes in the detergent formulation, such as amylase or mannanase, a protease inhibitor is often necessary to include in the formulation. An example of such an inhibitor is 4-formylboronic acid (4-FPBA) which has been used in the industry for more than 20 years. Such an inhibitor needs to bind at the active site of the protease to inhibit and stabilize the enzyme during storage in the detergent while being released upon dilution in the wash water to allow the protease to perform its cleaning function. Finding an alternative inhibitor or improving a known inhibitor as active ingredient with only laboratory testing is cumbersome, and resource intensive given the amount of candidate active ingredients that have to be tested. The invention described below in more detail allows to significantly improve this process.

[0042] One of the most commercially successful groups of crop-protection fungicides are the demethylase inhibitors (DMIs). These molecules derive their fungicidal efficacy from binding to, and inhibiting, fungal CYP51 enzymes also known as lanosterol-14-alpha- demethylases. Due to the agricultural and economic importance of compounds with this mode-of-action, they have been the focus of extensive research activities. For example, such efforts led to the discovery of mefentrifluconazole, the most recently commercialized DMI molecule.

[0043] Research targeting a new DMI fungicide often requires a search through many hundreds if not thousands of candidate molecules to simultaneously optimize several properties. These include avoidance of binding to similar enzymes in non-target organisms (e.g. mammals, plants, etc.), optimizing physico-chemical parameters such as solubility and vapor pressure to facilitate formulation and biological efficacy, and the avoidance of known unsuitable molecules.

[0044] However, while improving these various other properties it is essential that the changes to the molecular structure do not result in significant weakening of the binding to the CYP51 target enzyme. During the molecular design process, as ideas for new DMI molecules are created and prioritized, it would be advantageous to utilize computational approaches to assess the risk of this loss of target binding. However, currently available computation tools can often only be used to rather crudely assess a basic steric fit of any given molecule design in the CYP51 binding site, while the predicted binding energies are too inaccurate to have been useful in the discovery of mefentrifluconazole.

[0045] Unlike the conventional computational tools, the workflows of the invention described in more detail below, allow for a more advanced description of protein-ligand binding interactions and energies. Moreover, in this workflow semi-empirical quantum mechanics and density functional theory can be utilized for providing a more accurate description, for instance, of the key interaction of DMIs with the iron-haem cofactor. Thus, the invention has the potential to provide significantly improved predictions of CYP51 binding energies. Together with several other optimization parameters, these could enable a more efficient prioritization of molecular designs. This in turn can lead to a more effective use of synthesis resources and faster progress in product design, for instance, lead to products such as mefentrifluconazole in a faster and more efficient manner.

[0046] As described in the examples above and more generally shown in Fig. 1 , the interaction strength, for instance, represented by the binding energy between a target protein and a candidate active ingredient, for example, a candidate drug, is an important aspect in the development or improvement of active ingredients in a plurality of technical application contexts. This importance of the interaction strength is based on its correlation to the efficiency of the respective active ingredient, wherein the target efficiency for a drug in a technical application is often fundamental to the whole development process. For example, a drug that does not interact strongly enough with a target protein and is thus very ineffective might not be suitable for an application even if it provides theoretically a strong impact on the biological system that is targeted. On the other hand, an interaction that is too strong might also be unsuitable since in this case the candidate drug might not be selective enough, might also tend to bind to proteins that slightly differ from the target protein which in many applications is an undesired effect. Thus, the interaction energy provides a measure that can be utilized to decrease the huge amount of candidate active ingredients to a more reasonable amount that can be tested in laboratory screening with less resource assignment, since the actually tested candidate active ingredients with a respective higher reliability show the predetermined and suitable efficiency. This leads to much faster development procedures in the production of a final drug.

[0047] Fig. 2 shows schematically and exemplarily a method that is based on the above described principle to improve the active ingredient development process. Fig. 2 shows schematically and exemplarily more details of a method for testing active ingredients with respect to interaction strength with the target protein structure. The parts of the method before the performing of the test procedure refer to computer-implemented steps that can be realized on any form of computational environment, for example, on one or more computers, on a computer network or other distributed computing platform. The method starts with receiving digital twins of candidate active ingredients and a digital twin of a molecular cut-out of a target protein structure. For example, the digital twins can be received from a database on which respective digital twins are already stored and can be selected by an operator with respect to the predetermined technical application contexts. Fig. 5 shows schematically and exemplarily a data structure of respectively received digital twins and examples on how the digital twins can be realized in the digital domain. For example, a digital twin associated with a candidate active ingredient can be an ID of the candidate active ingredient, a SMILES representation, a SMARTS representation, a InChi representation or a Cartesian model providing Cartesian coordinates forthe molecules of the candidate active ingredient. A digital twin associated with the molecular cut-out of a target protein structure, for example, can be provided as a Cartesian model, or a geometric model as a digital twin in the digital domain. Utilizing a molecular cut-out that refers only to a part of the target protein structure that comprises the interaction site with the candidate active ingredient allows to focus the determination of the interaction strength on the important parts of the respective target protein. This allows to decrease the amount of utilized computational resources strongly, since in contrast to a complete protein that can comprise many thousands of atoms only a small but most important part with far less atoms, for instance, less than 1000, preferably, between 500 and 1000 atoms, has to be taken into account during the calculation. For the candidate active ingredients, a conformation search is then performed to determine one or more conformations associated with the one or more candidate active ingredients. Searching for conformations of the candidate active ingredients and then later utilizing the conformations for determining the energy measure allows for more reliable results when determining the respective interaction strength and thus also makes the testing procedure based on the interaction strength more reliable. The respective determined conformations can then be utilized for determining an energy measure for each of the conformations for the one or more candidate active ingredients. However, in an optional step, conformations can be selected from the determined conformations, for instance, based on the free energy of the conformations. For example, conformations that are most stable, i.e., are found in an energy minimum, can be selected and utilized in the energy measure determination. Further, the geometry of the respective conformations can be optimized before the energy measure determination.

[0048] Moreover, optional steps that increase the reliability of the determined interaction strength even more refer to determining from the candidate active ingredients and the molecular cut-out a respective molecular cluster referring to the candidate active ingredient interacting with the molecular cut-out at the interaction site. If a respective molecular cluster is determined, a conformation search for the candidate active ingredients interacting with the interaction site in the molecular cluster can also be performed. This is in particular advantageous since the interaction with the protein at the interaction site can influence the energy of the candidate active ingredient and thus can lead to other conformations, in particular, to other stable conformations as can be found for the free active ingredient. Thus, also in this embodiment, respective conformations, in particular, more stable conformations, can be selected. Moreover, a geometry optimization of the interaction site of the molecular cluster is performed optimizing the geometry of the interacting molecular cut-out with the interacting candidate active ingredient. Further, a geometry optimization of the molecular cut-out can also be performed.

[0049] Based on the determined and optionally selected and geometry-optimized conformations of the respective active ingredients and based on the preferably geometry-optimized molecular cut-out, an energy measure indicative of the interaction strength between the candidate active ingredient and the target protein structure can be determined. Optionally, the conformations can also be selected from the determined conformations in the interacting state for the candidate active ingredients with the geometry optimization of the molecular cluster. Preferably, the energy measure determination refers to determining a difference in an energy state between the candidate active ingredient and the molecular cut-out without any interaction and the molecular cluster representing the candidate active ingredient interacting with the molecular cut-out. For example, the energy measure can refer to a difference in a single-point gas phase energy, a translational, rotational and vibrational enthalpy / entropy and / or a difference in a solvation energy. Preferably, the above stated energy differences are added to determine a difference in the Gibbs free energy. The difference in the Gibbs free energy is indicative of the respective interaction strength between the candidate active ingredient and the molecular cut-out. However, also each of the above mentioned energies alone is already indicative of the interaction strength and can be utilized as energy measure. Based on the such determined energy measure, the digital twins associated with one or more candidate active ingredients can be provided for further testing with respect to one or more predetermined application goals. For example, the candidate active ingredients that should be utilized for further testing can be selected based on the determined energy measure by comparing the determined energy measure with a provided target energy measure range. Candidate active ingredients for which an energy measure is determined that falls within the target energy measure range can then be selected as test active ingredients. For such selected test active ingredients product data can be generated for controlling and / or monitoring a production of the test active ingredients, for instance, in a chemical production process.

[0050] Optionally, based on the produced test active ingredients, a respective test procedure can be performed with respect to one or more predetermined application goals. Such a test procedure can comprise producing the respective test active ingredients, subjecting the test active ingredients to respective experimental test procedures and validating the test procedure to generate a respective test result with respect to the one or more application goals. If the test result shows that a test active ingredient fulfils respective target technical application properties associated with the predetermined application goal, again production data can be generated for producing the active ingredient as final product or, for instance, in higher amounts for more extensive testing. If none of the tested test active ingredients fulfils a respective target technical application property, the method can again start with providing new candidate active ingredients. Since most of these procedures can be performed in an automated manner, it allows for a fast and resource-efficient screening for developing new and improved active ingredients in predetermined application contexts.

[0051] In the following, more detailed examples and embodiments of the general method above are provided. In an aspect, that is described in more detail as example in the following the invention relates to a workflow that combines efficient molecular conformer sampling tools with quantum mechanical tight-binding methods and density functional approximations for the calculation of Gibbs-free protein-drug interaction energies in a supramolecular clustering approach, including entropic contributions as well as solvation effects. In this example, the free energy of binding can be calculated as the difference between the free energies of the bound complex as molecular cluster, the target protein itself and the active ingredient. A visual representation of these molecular structures is provided by Fig. 3. Instead of considering the entire system of interest, for instance, the entire target protein, a molecular cut-out comprising the active interaction site harboring the active ingredient is explicitly calculated in order to shorten the calculation time. For example, a workflow as described above can be calculated in parallel on 20 CPUs with an AMD EPYC processor in to 3 to 5 hours.

[0052] The general workflow can comprise the following steps schematically shown in Fig. 4. First, the molecular cut-out of the binding site is generated by a radial section around the binding site, wherein different active ingredients can lead to different sections, due to different binding sites. The size of the cut-out, e.g. in angstroms, can also be varied if required. Since the geometry of the active ingredient can differ in the bound and unbound state, conformational sampling can be performed for the free active ingredient. Optionally, conformational sampling of the active ingredient interacting with the target protein, for instance, within the binding pocket, can be performed. Subsequently, the geometries of the generated molecular cluster, the molecular cut-out and the active ingredient are preferably optimized. In the case of the cluster and the cut-out, the C-alpha atoms, optionally also C, C and N, or all heavy atoms, can be constrained by a harmonic potential to prevent the cage-like structures from collapsing.

[0053] All three optimized structures can then form the starting point for further calculations. For example, the Gibbs free energies when the active ingredient binds to the protein can be calculated as the sum of three terms: Gbind= E+ GTRV+ GGSOIV.

[0054] On the left-hand-side, AE is the molecular electronic binding energy, GTRV the difference in translational, rotational, and vibrational enthalpy / entropy, and A<5GS0 / v the change in solvation energy transferring a molecule from the gas to the aqueous phase. The electronic binding energy refers to a difference in the electronic gas phase energy. For example, the electronic binding energy can be determined based on a difference of a single-point gas phase energy for the active ingredient, the molecular cut-out and the molecular cluster. Generally, a difference in the above equation refers to a difference between a) a quantity determined for the active ingredient plus the quantity determined for the molecular cut-out without binding and b) the quantity determined for the molecular cluster defined by the active ingredient binding to the molecular cut-out.

[0055] Fig. 5 shows exemplarily the method that can be utilized in the workflow. Preferably, for the method, in particular, the determination of the different energy measure contributions different methods with different levels of theory are utilized. For example, for conformational sampling, a conformer-rotamer ensemble sampling method, like CREST can be used in combination with a semiempirical extended tight-binding method, like GFN-xTB. CREST can be utilized for an in silico sampling of parts of the molecular low-energy chemical space by semi-empirical tight-binding methods combined with a meta-dynamics driven search algorithm. CREST applies an iterative meta-dynamics and genetic crossing (iMTD-GC) algorithm to generate molecular conformations. Preferably, for determining conformations at the binding site of the molecular cut-out the CREST method is utilized by constraining the binding site, for instance, the binding pocket, so that only conformations of the active ingredient at the constrained binding site are determined. This avoids also determining different conformations for the molecular cut-out itself. More detailed descriptions of these methods can be found, for example, in the articles “Automated exploration of the low- energy chemical space with fast quantum chemical methods.”, Pracht, P., F. Bohle, and S. Grimme, Physical Chemistry Chemical Physics, 2020. 22(14): p. 7169-7192 and “Extended tight-binding quantum chemistry methods.”, Bannwarth, C., et al., WIREs Computational Molecular Science, 2021. 11 (2): p. e1493, both incorporated herein by reference.

[0056] Geometry optimizations can be performed at both the GFN-FF and GFN2-xTB level of theory. GFN-FF and GFN2-xTB can again be used to calculate the harmonic vibrational frequencies. Enthalpic / entropic contributions can be derived within the modified rigid-rotor harmonic-oscillator approximation, for instance, as described in the article “Supramolecular Binding Thermodynamics by Dispersion-Corrected Density Functional Theory.”, Grimme, S., Chemistry - A European Journal, 2012. 18(32): p. 9955-9964 incorporated herein by reference. Single-point gas phase energies can be computed with different density functional approximations, for example, with r2SCAN-D4 and wB97X-D in a triple-zeta basis set, including dispersion corrections. Examples, for such calculations are given in the articles “r2SCAN-D4: Dispersion corrected meta-generalized gradient approximation for general chemical applications.”, Ehlert, S., et al., The Journal of Chemical Physics, 2021 . 154(6) and “Long-range corrected hybrid density functionals with damped atom-atom dispersion corrections.”, Chai, J.-D. and M. Head-Gordon, Physical Chemistry Chemical Physics, 2008. 10(44): p. 6615-6620 both incorporated herein by reference. Solvation effects at the semi-empirical level can be included by a GBSA and ALPB model, at the DFT level COSMO can be applied. Examples, for these methods are described, for instance, in the article, “Robust and Efficient Implicit Solvation Model for Fast Semiempirical Methods. ” Ehlert, S., et al., Journal of Chemical Theory and Computation, 2021 . 17(7): p. 4250-4261 , and “The COSMO and COSMO-RS solvation models.”, Klamt, A., WIREs Computational Molecular Science, 2011. 1 (5): p. 699-709 all incorporated herein by reference. Combined, this leads to the workflow shown in Fig. 5.

[0057] Figs. 7 and 8 show examples of free energies determined with the screening method compared to literature values. The Figs. 7 and 8 show in the upper left corner noted with MM / GBSA results from literature for current force-field based standard calculations. These values are derived from “Accurate and Reliable Prediction of Relative Ligand Binding Potency in Prospective Drug Discovery by Way of a Modern Free-Energy Calculation Protocol and Force Field”, Wang et. Al., Journal of the American Chemical Society 2015 137 (7), 2695-2703. The other three shown graphs show free energies derived utilizing the above described method, wherein Ievel2 means utilizing SQM (semi-empirics), levels means utilizing DFT, radius refers to the utilized cut-out radius in Angstrom, and fixed heavy atoms means that during a geometric optimization all atoms but hydrogen are constrained. Fig. 7 shows respective free energy values for 16 active ingredients binding to the nonreceptor tyrosine-protein kinase TYK2 and Fig. 8 shows 25 active ingredients binding to the induced myeloid leukemia cell differentiation protein MCL1 . All graphs show the above described method allows for a reliable determination of respective effective binding of the respective active ingredients to the target protein and further outperforms the currently utilized standard method.

[0058] A furthertechnical application example refers to the above already mentioned demethylase inhibitors (DMIs), e.g., Revysol® / mefentrifluconazole, a commercially very successful group of systemic fungicides that target cell membrane integrity by inhibiting C14 demethylation during sterol formation. DMIs bind and inhibit fungal CYP51 , a metalloenzyme that contains heme as a cofactor and is a coordination complex consisting of an iron ion coordinated to a tetrapyrrole, which acts as a tetradentate ligand, and one or two axial ligands. Utilizing the state of the art methods the interaction of active ingredients with the open-shell Fe(lll) low-spin ion is difficult to compute and most common force-fields, if not specially parameterized, fail to describe the interaction correctly. In the following a test set of eight potential DMIs is provided and the results of the above described method are compared with the results of standard MM / GBSA models with experimentally measured binding free energies. The basic structure of the eight molecules is shown below, where the unit coordinating the iron (indicated by the R) is mutated.

[0059] The results are shown in Figs. 9a to 9e. Fig. 9a shows the results of the standard MM / GBSA method, Fig. 9b shows the results of the above described method at level 2 meaning AE, AGsoiv and AGTRV are computed with a GFN-xTB method, a constrain on C-alpha, and a cut-out radius of 6 Angstrom. Fig. 9c shows level 3 results, meaning AE is computed with a DFT method, for instance, r2SCAN / def2-TZVP, and AGsoiv is obtained with a COSMO method and AGTRV is obtained with a GFN-xTB method, a constrain on C-alpha, and a cutout radius of 5 Angstrom. The correlation R2value is significantly increased by the above described method from 0.01 for MM / GBSA to 0.41 for level 2 and 0.46 for level 3, respectively. To understand the improved performance of the above described method over MM / GBSA, the binding free energy is decomposed into its components, namely the electronic binding energy plus solvation free energy AE+ AGsoiv is solved and the entropic contributions AGTR computed in a modified rigid-rotor harmonic-oscillator approximation. The results are shown in Fig. 9d for the AE and AGsoiv contribution to the level 2 result as shown in Fig. 9b, and Fig 9e for the AGTRV contribution to the level 2 result as sown in Fig. 9b. It can be seen that in this example neitherthe electronic energy embedded in the GBSA model with R2= 0.12 nor the entropic contributions with R2= 0.09 alone represent an improvement over MM / GBSA. But, the sum of all energy contributions leads in this example to a significant improvement compared to the standard force field-based MM / GBSA model. However, in other examples also the contributions alone can be improved.

[0060] Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.

[0061] For the processes and methods disclosed herein, the operations performed in the processes and methods may be implemented in differing order. Furthermore, the outlined operations are only provided as examples, and some of the operations may be optional, combined into fewer steps and operations, supplemented with further operations, or expanded into additional operations without detracting from the essence of the disclosed embodiments.

[0062] In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single unit or device may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0063] Procedures like the receiving of a digital twin associated with a respective candidate active ingredient, receiving a digital twin associated with a molecular cut-out of a target protein structure, determine one or more conformations determining an energy measure, providing digital twins associated with one or more of the candidate active ingredients, can be performed by any other number of units or devices. These procedures can be implemented as program code means of a computer program and / or as dedicated hardware.

[0064] A computer program product may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0065] Any units described herein may be processing units that are part of a classical computing system. Processing units may include a general-purpose processor and may also include a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or any other specialized circuit. Any memory may be a physical system memory, which may be volatile, non-volatile, or some combination of the two. The term “memory” may include any computer-readable storage media such as a non-volatile mass storage. If the computing system is distributed, the processing and / or memory capability may be distributed as well. The computing system may include multiple structures as “executable components”. The term “executable component” is a structure well understood in the field of computing as being a structure that can be software, hardware, or a combination thereof. For instance, when implemented in software, one of ordinary skill in the art would understand that the structure of an executable component may include software objects, routines, methods, and so forth, that may be executed on the computing system. This may include both an executable component in the heap of a computing system, or on computer- readable storage media. The structure of the executable component may exist on a computer-readable medium such that, when interpreted by one or more processors of a computing system, e.g., by a processor thread, the computing system is caused to perform a function. Such structure may be computer readable directly by the processors, for instance, as is the case if the executable component were binary, or it may be structured to be interpretable and / or compiled, for instance, whether in a single stage or in multiple stages, so as to generate such binary that is directly interpretable by the processors. In other instances, structures may be hard coded or hard wired logic gates, that are implemented exclusively or near-exclusively in hardware, such as within a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or any other specialized circuit. Accordingly, the term “executable component” is a term for a structure that is well understood by those of ordinary skill in the art of computing, whether implemented in software, hardware, or a combination. Any embodiments herein are described with reference to acts that are performed by one or more processing units of the computing system. If such acts are implemented in software, one or more processors direct the operation of the computing system in response to having executed computerexecutable instructions that constitute an executable component. Computing system may also contain communication channels that allow the computing system to communicate with other computing systems over, for example, network. A “network” is defined as one or more data links that enable the transport of electronic data between computing systems and / or modules and / or other electronic devices. When information is transferred or provided over a network or another communications connection, for example, either hardwired, wireless, or a combination of hardwired or wireless, to a computing system, the computing system properly views the connection as a transmission medium. Transmission media can include a network and / or data links which can be used to carry desired program code means in the form of computer-executable instructions or data structures and which can be accessed by a general-purpose or special-purpose computing system or combinations. While not all computing systems require a user interface, in some embodiments, the computing system includes a user interface system for use in interfacing with a user. User interfaces act as input or output mechanism to users for instance via displays.

[0066] Those skilled in the art will appreciate that at least parts of the invention may be practiced in network computing environments with many types of computing system configurations, including, personal computers, desktop computers, laptop computers, message processors, hand-held devices, multi-processor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, mobile telephones, PDAs, pagers, routers, switches, datacenters, wearables, such as glasses, and the like. The invention may also be practiced in distributed system environments where local and remote computing system, which are linked, for example, either by hardwired data links, wireless data links, or by a combination of hardwired and wireless data links, through a network, both perform tasks. In a distributed system environment, program modules may be located in both local and remote memory storage devices. Those skilled in the art will also appreciate that at least parts of the invention may be practiced in a cloud computing environment. Cloud computing environments may be distributed, although this is not required. When distributed, cloud computing environments may be distributed internationally within an organization and / or have components possessed across multiple organizations. In this description and the following claims, “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources, e.g., networks, servers, storage, applications, and services. The definition of “cloud computing” is not limited to any of the other numerous advantages that can be obtained from such a model when deployed. The computing systems of the figures include various components orfunctional blocks that may implement the various embodiments disclosed herein as explained. The various components or functional blocks may be implemented on a local computing system or may be implemented on a distributed computing system that includes elements resident in the cloud or that implement aspects of cloud computing. The various components or functional blocks may be implemented as software, hardware, or a combination of software and hardware. The computing systems shown in the figures may include more or less than the components illustrated in the figures and some of the components may be combined as circumstances warrant.

[0067] Any reference signs in the claims should not be construed as limiting the scope.

[0068] The invention refers to a computer implemented method for testing active ingredients with respect to an interaction strength with a target protein structure. For candidate active ingredients a digital twin is received. A digital twin associated with a molecular cut-out of a target protein structure is received. Conformations associated with the candidate active ingredients are determined based on the digital twin associated with the respective candidate active ingredients. An energy measure for the candidate active ingredients is determined based on the conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the interaction site with the target protein structure. Based on the determined energy measure, digital twins associated with one or more of the candidate active ingredients for further testing with respect to one or more application goals.

Claims

Claims:1 . Computer implemented method for testing active ingredients with respect to an interaction strength with a target protein structure, wherein the method comprises: receiving for one or more candidate active ingredients a digital twin associated with a respective candidate active ingredient, receiving a digital twin associated with a molecular cut-out of a target protein structure, wherein the molecular cut-out comprises an interaction site between the target protein structure and an active ingredient, determine one or more conformations associated with the one or more candidate active ingredients based on the digital twin associated with the respective one or more candidate active ingredients, determining an energy measure for the one or more candidate active ingredients based on the one or more conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the interaction site with the target protein structure, providing, based on the determined energy measure, digital twins associated with one or more of the candidate active ingredients for further testing with respect to one or more application goals.

2. The method according to claim 1 , wherein the molecular cut-out is generated by cutting the target protein molecular structure in a predetermined radius around to the binding site.

3. The method according to any of claims 1 and 2, wherein the determining of one or more conformations comprises, selecting one or more conformations based on a free energy of the conformations associated with a candidate active ingredient.

4. The method according to any of the preceding claims, wherein the method further comprises determining one or more conformations associated with one or more active ingredients interacting at the interaction site with the molecular cut-out based on the digital twin associated with respective one or more candidate active ingredients and the digital twin associated with the molecular cut-out, wherein the energy measure is furtherdetermined based on the one or more conformations associated with one or more active ingredients interacting at the interaction site with the molecular cut-out.

5. The method according to any of the preceding claims, wherein the energy measure for a conformation of an active ingredient is determined based on an energy difference between an energy of the conformation and the molecular cut-out without interaction and an energy of a respective target protein cluster, wherein the target protein cluster is associated with a conformation interacting with the interaction site of the molecular cut-out.

6. The method according to any of the preceding claims, wherein the energy measure for a conformation of an active ingredient is determined based on at least one of electronic binding energy; a single-point gas phase energy; a translational, rotational, and / or vibrational enthalpy; a translational, rotational, and / or vibrational entropy; and a solvation energy.

7. The method according to any of the preceding claims, wherein the method further comprises optimizing a geometry of the molecular cut-out and / or a target protein cluster, wherein the target protein cluster is associated with a conformation interacting with the binding site of the molecular cut-out, wherein the energy measure is determined based on the digital twin of the respective optimized molecular cut-out and / or the target protein cluster.

8. The method according to claim 7, wherein for optimizing of the geometry of the molecular cut-out and / or the target protein cluster the positions of atoms belonging to at least one atom class are constraint.

9. The method according to any of the preceding claims, wherein the receiving of the digital twin of the molecular cut-out, comprises completing the molecular cut-out by adding predetermined atoms or atom groups to bonds that have been cut.

10. The method according to any of the preceding claims, wherein the energy measure is the Gibbs free energy.

11. A test method for testing active ingredients with respect to one or more application goals, wherein the method comprises:performing a method according to any of the preceding claims for determining one or more active ingredients with a predetermined interaction strength with a target protein structure, producing the one or more active ingredients based on the provided digital twins of the one or more active ingredients, subjecting the produced one or more active ingredients to further testing to determine an applicability of the one or more active ingredients with respect to one or more application goals.

12. An apparatus for testing active ingredients with respect to an interaction strength with a target protein structure, wherein the apparatus comprises one or more processors configured to perform a method comprising receiving a digital twin associated with one or more candidate active ingredients, receiving a digital twin associated with a molecular cut-out of a target protein structure, wherein the molecular cut-out comprises an interaction site between the target protein structure and an active ingredient, determine one or more conformations associated with one or more candidate active ingredients based on the digital twin associated with the respective one or more candidate active ingredients, determining an energy measure for the one or more candidate active ingredients based on the one or more conformations and the digital twin associated with the molecular cut-out, wherein the energy measure is indicative of the interaction strength between an active ingredient interacting at the binding site with the target protein structure, providing, based on the determined energy measure, digital twins for one or more active ingredients of the candidate active ingredients for further testing with respect to one or more application goals.

13. A test system for testing active ingredients to with respect to one or more application goals, wherein the system comprises:an apparatus according to claims 1 1 for performing a method according to any of claims 1 to 9 for determining one or more active ingredients with a predetermined interaction strength with a target protein structure, a production facility configured for producing the one or more active ingredients based on the provided digital twins of the one or more active ingredients, a testing facility configured for subjecting the produced one or more active ingredients to further testing to determine an applicability of the one or more active ingredients with respect to one or more application goals.

14. A computer program product for testing active ingredients, wherein the computer program product causes an apparatus according to claims 12 to carry out the method according to any of claims 1 to 9 when executed by the apparatus.

15. A computer program product for testing active ingredients, wherein the computer program product causes a system according to claims 13 to carry out the method according to claim 11 when executed by the system.

Images