Method for determining the biodegradability of functional compounds

Abstract

The present invention relates to a method for determining the biodegradability of a functional compound. A digital representation of the functional compound is provided, indicating the physicochemical properties of the functional compound. Additionally, habitats are provided, indicating habitat descriptor values ​​that affect the biodegradation of the functional compound. The habitat descriptors indicate environmental characteristics of the habitats. A biodegradation model is provided based on the habitats, and the biodegradation model is adapted to determine the biodegradability of the functional compound in each habitat, where the biodegradation model is a data-driven model parameterized with respect to the habitat to determine the biodegradability of the functional compound based on the physicochemical properties. The biodegradation of the functional compound is then determined based on the provided biodegradation model and the digital representation of the functional compound.

Description

[Technical Field]

[0001] FIELD OF THE INVENTION The present invention relates to a method, an apparatus and a computer program product for determining biodegradability that can be used to verify the biodegradation of functionalized compounds. The present invention also refers to a training method, a training apparatus and a training computer program for training a data-driven biodegradation model that can be used by the method, apparatus and computer program product for determining the biodegradability of functionalized compounds. The present invention also refers to a method and an apparatus for providing an interface for determining the biodegradability of functionalized compounds. [Background technology]

[0002] Background of the Invention Functional compounds, generally referring to small molecules, are widely used in industrial and / or everyday products due to their wide range of application properties. The uses of functional compounds include, among others, excipients, plasticizers, stabilizers, inhibitors, odorants, fragrances, nutrients, catalysts, radiation absorbers, lubricants, and surfactants. However, this wide range of applications results in a huge amount of waste containing the functional compounds used. Non-degradable waste poses problems when disposed of in undesignated environments. In particular, the accumulation of chemicals in the environment, such as the accumulation of phosphates that leads to algae blooms, is undesirable. Therefore, not only are functional compounds that decompose necessary, but knowledge of the biodegradability of functional compounds must also be considered early in the product design process. Therefore, it would be advantageous to provide a possibility to predict the biodegradability of functional compounds accurately and computationally inexpensively. Summary of the Invention [Problem to be solved by the invention]

[0003] Summary of the Invention It is an object of the present invention to provide a method, an apparatus and a computer program product that allows for accurate determination of the biodegradability of functional compounds, which is computationally inexpensive and can be robustly applied to new functional compounds. It is also an object of the present invention to provide a training method, a training apparatus and a computer program product that can be used in the method, the apparatus and the computer program product and that can provide a biodegradation model that can be trained to provide good determination accuracy by utilizing fewer computational resources. [Means for solving the problem]

[0004] In a first aspect of the present invention, a computer-implemented method for determining biodegradability that can be used to verify the biodegradation of a functional compound is provided, the method comprising: i) providing a digital representation of the functional compound; ii) providing biodegradation habitats, the biodegradation habitats indicating habitat description values ​​of habitat descriptors that affect the biodegradation of the functional compound in each habitat, the habitat descriptors indicating environmental characteristics of the habitat; iii) providing a biodegradation model based on the provided biodegradation habitats, the biodegradation model being adapted to determine the biodegradability of the functional compound in each biodegradation habitat, the biodegradation model being a data-driven model parameterized with respect to the biodegradation habitat to determine the biodegradability of the functional compound based on the digital representation of the functional compound; and iv) determining the biodegradability of the functional compound based on the provided biodegradation model and the digital representation of the functional compound.

[0005] The biodegradation model is specifically adapted to determine the biodegradability of a functional compound in a given biodegradation habitat, characterized by respective habitat descriptor values ​​that affect the biodegradation of the functional compound in each habitat, allowing for highly accurate determination of the biodegradability of the functional compound in each habitat. Furthermore, because the biodegradation model is specifically trained for one or more specific biodegradation habitats, less training data is required, making the biodegradation model more flexible with regard to determining the biodegradability of new functional compounds that are not part of the training data set. This method therefore allows for accurate determination of biodegradability that is computationally inexpensive and can be flexibly applied to new functional compounds. Furthermore, whereas traditional testing methods for testing the biodegradability of functional compounds are very time-consuming and can take months or years to obtain results, the above-described method can provide results essentially immediately. This not only reduces the technical requirements for determining biodegradability, but also significantly shortens the time required to design new biodegradable products. Furthermore, by making it easy to consider accurate determination of a product's biodegradability already during the design process, it becomes possible to design products that avoid bioaccumulation. In particular, it can be ensured that the functional compounds used in the products are biodegradable in the respective expected environments, for example marine habitats.

[0006] The development of new chemical products tailored to application requirements is a major challenge in the modern chemical industry. In recent years, additional requirements have arisen regarding the environmental impact of chemical products along their life cycle. One important aspect of environmental impact is the prevention of bioactive residues and bioaccumulation. Bioactive residues, i.e., residues that affect the flora and / or fauna of the habitat, are an increasing problem and can be avoided if the functional compounds are biodegradable. Bioaccumulation, the gradual accumulation of compounds in organisms, can be avoided if the functional compounds are biodegradable. A series of standardized tests are currently used to evaluate biodegradation. Various tests exist for biodegradability using specific conditions (e.g., ISO 13432, ISO 14852, ISO 14855, ISO 17556, and OECD 301). Standardized tests often balance time-efficient testing (minimum 14 days, maximum 24 months) with real-life conditions. In fact, to speed up testing, temperatures higher than those in real life conditions are often used. Companies developing new functional compounds must invest significant resources in self-assessment and certification of the sustainability of their products. The entire biodegradability assessment, including laboratory space and equipment, is costly and time-consuming. Therefore, the biodegradability of new functional compounds needs to be identified early in the development process. The proposed method for determining biodegradability disclosed herein enables faster and more efficient development of new functional compounds. Biodegradability can be determined at an early stage, even before the functional compound is synthesized. This makes it possible to determine whether the functional compound is suitable for market entry, thereby shortening the time to market. Since there is no need to synthesize the functional compound to determine its biodegradability, waste generation can also be reduced. The proposed method provides a digital twin of the measurement of the biodegradability of a functional compound.

[0007] Furthermore, standard measurements and tests of biodegradability are often time-consuming, including waiting times of up to several months or even years. Specifically, when developing new functional compounds for each application, these time-consuming tests can severely limit the development process. In this regard, the present invention makes it possible to provide results for new functional compounds immediately, significantly reducing the time before results are available.

[0008] Furthermore, because the number of potentially suitable functional compounds for a particular application is very large and, in many cases, many of these functional compounds have not been thoroughly researched, technical product engineers today are faced with the technical challenge of finding functional compounds that are not only suitable for a particular application but also satisfy the respective target properties. To find each functional compound that may be particularly suitable for the application, they must synthesize and test a huge number of possible functional compounds or examine vast data sets and libraries in which potential functional compounds are stored. Even when using sophisticated experimental design, it is still necessary to synthesize and experimentally test a very large number of possible functional compounds. In this regard, the above-described method can help users, such as technical product engineers, automatically find potentially suitable functional compounds much more quickly. In particular, by using the above-described method, users only need to synthesize and test potentially suitable functional compounds that are determined to be highly likely to satisfy the respective target properties, particularly the target biodegradability. Therefore, unnecessary synthesis and testing of functional compounds can be avoided. Therefore, the present method allows users to perform the technical task of finding functional compounds suitable for technical applications quickly and efficiently.

[0009] The method relates to a computer-implemented method and can therefore be carried out by a general-purpose or special-purpose computer adapted to carry out the method, for example by executing a respective computer program. The method is adapted to determine the biodegradability of a given functional compound.

[0010] A biodegradable functional compound refers to a functional compound that can be decomposed by biological processes. In particular, a biodegradable functional compound can refer to a functional compound that can be assimilated by bacteria and / or fungi to produce environmentally friendly products, i.e., to decompose into non-polluting residues. Generally, the determined biodegradability can refer to any quantification of the biodegradability of a functional compound. For example, the determined biodegradability can refer to only one value (e.g., the half-life of the functional compound in each habitat) or multiple values, such as the time-dependent degradation function of the functional compound in a particular habitat. Preferably, biodegradability refers to the percentage value of biodegradation after a predetermined time frame. Generally, biodegradability refers to a measure of the degradation, or decomposition, of a functional compound caused by a biological process, i.e., a process involving biological materials, particularly microorganisms, involved in the degradation process. Therefore, biodegradability does not refer to a purely chemical decomposition process that does not involve microbial activity. Biodegradability is an inherent property of a functional compound. In this context, the intrinsic properties of a functional compound refer to properties of the functional compound that, in relation to a particular context, are caused by and thus reflect the properties of the functional compound, i.e., its structure, composition, etc. In particular, biodegradability reflects the properties of the functional compound when present in a particular biologically active environment. For example, the biodegradability of a functional compound preferably refers to any one of the mineralization properties, biotransformation properties, and / or degradation of the functional compound after a particular time frame. Biodegradability can be used, in particular, to verify the biodegradation, i.e., degradation properties, of a functional compound in, for example, a particular biodegradation environment, i.e., biodegradation habitat. For example, biodegradability can be compared with the biodegradability required for a particular application, thereby determining whether a particular functional compound is suitable for that application.

[0011] Biodegradable compounds can be designed to break down through the action of living organisms upon disposal. Biodegradability can relate to the environmental fate and / or behavior of a compound. Biodegradability can relate to the extent to which a compound can be broken down by microorganisms such as bacteria, fungi, or algae. Biodegradability can depend on the chemical structure, molecular weight, physical factors such as crosslink density, branching, crystallinity, or solubility of the compound's components, and exposure conditions, such as habitats such as soil, compost, or aquatic systems. With respect to exposure conditions, substrate properties such as microorganisms, microbial populations, nutrient concentrations, temperature, pH, pO2, ionic conditions, or toxicity affect biodegradability. Biodegradability can be measured based on measured mass loss (mg / hour), dissolved organic carbon (DOC, organic carbon concentration / hour), oxygen consumption (e.g., pressure measurements, e.g., Pa / hour), or carbon dioxide production over time (e.g., pressure measurements, e.g., Pa / hour).

[0012] Many measurement standards have been developed to quantify biodegradability in terms of measured properties of compounds. Various measurement methods have been defined to determine biodegradability under predetermined laboratory conditions. For example, six methods for determining biodegradability are described in "Readily Degradable" (July 17, 1992) for wastewater OECD Test 301. Furthermore, for example, ASTM D5988-18, "Standard Test Method for Determining the Aerobic Biodegradation of Plastic Materials in Soil," describes the measurement of carbon dioxide evolved by microorganisms as a function of exposure time, thus determining the degree of biodegradation compared to a reference material. Furthermore, for example, ISO 17556:2019, "Plastics - Determination of the Ultimate Aerobic Biodegradation in Soil by Measuring Oxygen Consumption or Carbon Dioxide Evolution Using a Respirometer," provides the optimal biodegradation rate of plastic materials in test soil by controlling oxygen consumption or carbon dioxide production. Further, for example, ISO 14855-1:2012, "Determination of the ultimate aerobic biodegradability of materials under controlled composting conditions - Method by measuring carbon dioxide evolution - Part 1: General method" and ASTM D5338-15, "Standard test method for determining the aerobic biodegradation of plastic materials under controlled composting conditions, including thermophilic temperatures," determine the ultimate aerobic biodegradability (meaning that microorganisms completely consume the chemical or organic material in the presence of oxygen) of organic compound-based plastics under controlled composting conditions by measuring the percent conversion of carbon dioxide to carbon dioxide and the extent of plastic breakdown at the end of the test. ASTM D6400-21, "Standard Specification for Labeling of Plastics Designed to be Aerobically Composted in Municipal or Industrial Facilities," additionally includes elemental analysis, plant germination (phytotoxicity), and mesh filtration of the resulting particles. ISO 17088:2021, "Plastics -- Organic recycling -- Specification for compostable plastics," includes an assessment of adverse effects on the composting process and equipment, as well as on the quality of the resulting compost, including the presence of high levels of restricted metals and other harmful components.

[0013] Regarding aerobic biodegradation, ISO 18830:2016 "Determination of aerobic biodegradation of non-floating plastic materials at the plastic-seawater / sandy sediment interface - Method by measuring oxygen demand with a closed respirometer" and ISO 19679:2020 "Determination of the degree of aerobic biodegradation of non-floating plastic materials at the plastic-seawater / sediment interface - Method by analysis of carbon dioxide evolved" have been created. Biodegradation assessment is measured by oxygen demand or CO2 evolved. Further standards include, for example, ISO 14853:2016 "Plastics -- Determination of anaerobic biodegradation in aqueous systems -- Method by measurement of biogas production", ISO 23977-1:2020 "Plastics -- Determination of aerobic biodegradation of plastic materials exposed to seawater -- Part 1: Method by analysis of carbon dioxide evolved" and ISO 23977-2:2020 "Plastics -- Determination of aerobic biodegradation of plastic materials exposed to seawater -- Part 2: Method by measurement of oxygen demand with a closed respirometer".

[0014] The quantified biodegradation properties of a compound may depend on the measurement method and conditions used, the measurement environment, and measurements related to the degradation process, such as mass loss, DOC, oxygen consumption, or carbon dioxide production over time. The measurement method and measured properties may be provided as metadata for each measurement point related to biodegradation.

[0015] Generally, functional compounds can be any functional compound. Functional compounds generally have application properties for technical purposes, i.e., perform a function in chemical products. For example, the compound that provides UV protection in sunscreens is a functional compound. Functional compounds can include active ingredients, i.e., components that provide the functionality of the functional compound. Additionally, active ingredients can provide the biological activity of the functional compound. For example, active ingredients can refer to antifungal agents, fragrance chemicals, UV absorbers, food additives, vitamins, nutrients, dyes, and surfactants. Functional compounds are generally characterized by their chemical structure. However, different chemical structures can exist within a single functional compound. For example, functional compounds can be composed of components that are molecules that undergo tautomerization, protonation, or deprotonation, etc. Functional compounds can be composed of two or more stereoisomers. Thus, functional compounds that constitute one type of molecule can be associated with one or more chemical structures, such as one protonated structure and one uncharged structure, or two stereoisomers. Thus, in general, a functional compound can include all molecules related to a single chemical formula through isomerization, such as (de)protonation, tautomerization, and stereoisomerism. Furthermore, a functional compound can refer to any functional compound that can be described by one or more chemical structures. In one embodiment, the chemical structures can be related to a single chemical formula.

[0016] Preferably, the functional compound consists of a small molecule. Preferably, the functional compound has a molecular mass of less than 10,000 g / mol. More preferably, the compound has a molecular weight of less than 800 g / mol, even more preferably less than 400 g / mol. Furthermore, the functional compound preferably exists in the environment in a form that allows the molecule of the functional compound to be completely described using a simple structural formula containing relevant information. A simple molecular structure refers to a molecule that can be clearly described by the covalent bonds between the atoms of the molecule. Examples of cases where this is not the case are systems that have a dynamic equilibrium between several forms, such as monomers and oligomers, as in the case of some inorganic acids, or ionic species with highly localized charges that strongly interact with the solvent, for example, via hydrogen bonds. Preferably, the functional compound has at least one of the following properties: it has an effect on a living organism, is suitable for influencing the structure, or is suitable for influencing the function of a living organism. In one embodiment, the functional compound comprises at least one of the following functional groups: an ether group, a hydroxyl group, a peroxo group, a hydroperoxide group, a carboxyl group, a carboxyl group derivative, a carbonyl group, an amine group, an imine group, a hydrazine group, a urea group, a urethane group, a thiourethane group, a nitrile group, an azide group, an azo group, a cyanate group, an isocyanate group, an isocyanide group, a pyridine group, an alkane group, an alkene group, an alkyne group, a phenyl group, a ketone group, a thioketone group, an aldehyde group, a thioaldehyde group, an acetal group, a ketal group, an oxime group, a hydrazone group, a nitro group, a nitroso group, a thiol group, a sulfide group, a disulfide group, a sulfonic acid derivative, a sulfinic acid derivative, a sulfate group, a sulfate derivative, a sulfone group, a sulfoxide group, a sulfhydryl group, a sulfide group, a phosphorane group, a phosphate group, or a phosphate derivative. phosphonates, phosphine groups, silane groups, silazane groups, silicone groups, borate groups, borane groups, halide groups or combinations thereof. Preferably, the functional compounds correspond to the following compound classes: carboxyl derivatives, ether groups, amine groups, hydroxy groups, carbonyl groups, alkane groups, alkene groups, benzene derivatives, pyridine derivatives, halide groups.

[0017] In a first step, the method includes providing a digital representation of the functional compound. Specifically, providing may refer to receiving the digital representation from a user's input, for example, using a respective input unit. Additionally, providing may refer to accessing a storage unit in which the digital representation is already stored. The digital representation of the functional compound may be any representation that defines the functional compound and / or provides information that allows deriving respective characterizing parameters, e.g., physicochemical properties of the functional compound. Preferably, the digital representation includes instructions for characterizing parameters that refer to the chemical structure and / or synthetic specifications of the functional compound. Physiochemical properties may optionally be derived and used based on the digital representation.

[0018] In a preferred embodiment, the digital representation is a chemical structure and / or synthesis specification of the functional compound. As mentioned above, the functional compound may also include one or more chemical structures. In this case, the digital representation preferably provides a chemical structure that relates to and / or indicates one or more structural formulas and the quantitative ratio of two or more structural formulas present in the functional compound. Preferably, at least two structural formulas provided by the digital representation for the functional compound correspond to structural formulas related via chemical equilibrium. Furthermore, if at least one synthesis specification is known for the functional compound, it is preferable to provide the synthesis specification as part of the digital representation. However, in some cases, the target functional compound is determined first, and then the synthesis specification of the target functional compound is determined. This method can be particularly utilized for potential target compounds whose synthesis specifications are not yet known.

[0019] Because the structural formula of a functional compound can be influenced by its habitat, the chemical structure provided by the digital representation preferably depends on the habitat. For example, in aqueous media, a particular molecule may tend to exist in a protonated form with a 1:3 unprotonated-to-protonated ratio. In this case, a digital representation of such a molecule using a single chemical structure may not be sufficient. Such a molecule may be represented by a digital representation that includes a quantitative ratio indicating the equilibrium between different structures associated with a single chemical formula, where two or more structures are in chemical equilibrium. The digital representation may be referred to as a statistical representation because it incorporates the statistical frequency of molecules associated with each structure. To illustrate the concept, an example equilibrium between morpholine and one of its protonated structures is illustrated: morpholine + H+ (25%) ⇔ protonated morpholine (75%). Both morpholine and protonated morpholine can be the result of introducing morpholine into water at a particular pH value that tends toward protonated morpholine; for example, the ratio can be 1:3 protonated morpholine to morpholine. Thus, a digital representation of introducing morpholine into water can refer to a structural specification of morpholine and protonated morpholine, including the respective amounts or ratios of amounts of two or more structural formulae present in the functional compound. An example of a chemical structure specification can be the number and types of atoms and their respective connectivity. Another example includes using SMILE and / or SMARTS to represent the chemical structure of a functional compound.

[0020] The digital representation can also be a chemical graph of the functional compound. A chemical graph is a representation of the structural formula of a compound from the perspective of graph theory. For example, a chemical graph can be a labeled graph whose vertices correspond to the atoms of the compound and whose edges correspond to chemical bonds. The digital representation can then refer to a graph in which the atoms of the molecule are the nodes and the atomic bonds of the molecule are the edges of the graph.

[0021] Preferably, the digital representation includes parameters characterizing the functional compound, in particular the physicochemical properties. In particular, the physicochemical properties of the functional compound can be quantified by physicochemical parameters. Preferably, the digital representation indicates and / or includes physicochemical parameters, preferably referring to respective descriptors, which indicate the physicochemical properties of the functional compound. In particular, the physicochemical parameters indicate parameters that quantify the physicochemical properties of the functional compound. In this context, the term "physicochemical properties" refers to the physical and / or chemical properties of the functional compound. However, the digital representation can also be provided to allow, for example, the physicochemical properties to be derived, for example, by providing a representation of the functional compound in which the respective physicochemical properties are already stored or can be determined, for example, by respective calculations. Preferably, the digital representation indicates at least one of the synthetic specifications, structural formula, brand name, IUPAC name, chemical identifier, and CAS number of the functional compound.

[0022] Preferably, the physicochemical parameters refer to at least one of a compositional descriptor, a count descriptor, a list of structural fragments, a fingerprint, a graph invariant, a 3D descriptor, and / or a higher-dimensional descriptor, which represent parameters quantifying the physicochemical properties of the functional compound. In a preferred embodiment, the functional compound descriptor refers to a 3D descriptor, specifically a quantum chemical descriptor. Furthermore, the inventors have found that the molar mass in particular describes the biodegradation of functional compounds very accurately. Therefore, it is particularly preferred that the physicochemical parameters include the molar mass of the functional compound. Possible physicochemical parameters are defined in more detail below.

[0023] The constituent descriptors may refer to any of the potential, molecular weight, charge, spin, boiling point, melting point, enthalpy of fusion, dissociation constant, Hansen parameters, protic, polar and dispersive contributions, Abraham parameters, retention index, total polar surface area, receptor binding constant, Michaelis-Menten constant, inhibitor constant, mutagenicity, LD50, bioconcentration, toxicity, biodegradation profile and viscosity.

[0024] The count descriptor may refer to any of the sum of atomic electronegativities, sum of atomic polarizabilities, number of atoms and non-hydrogen atoms, number of H, B, C, N, O, P, S, Hal and heavy atoms, number of hydrogen donor and hydrogen acceptor atoms, number of bonds, number of non-hydrogen or multiple bonds, number of double bonds, triple bonds and aromatic bonds, number of functional groups, functional group ratio, sum of bond orders, aromatic ratio, number of rings or cycles, number of unpaired electrons, number of rotatable bonds, rotatable bond fraction and number of conformers.

[0025] The physicochemical parameter referring to the list of structural fragment descriptors may refer to at least one of a list of molecular fractions, a list of functional groups, a list of bonds, and a list of atoms. The fingerprint descriptor preferably includes at least one of a MACCS key in bit format or total amount format, a Morgan Fingerprint and other circular fingerprints, preferably in bit format or total amount format, a topological twist, an atom pair, an infrared spectrum and related spectra, a fingerprint number, a PubChem fingerprint, a substructure fingerprint, and a Klekota-Roth fingerprint. The graph invariance / topology index descriptor preferably includes at least one of a topostructural index and a topochemical index. Furthermore, molecular graph representations can be used as descriptors.

[0026] In a preferred embodiment, the physicochemical parameters of the functional compound are 3D descriptors comprising at least one of the following: molecular volume, average volume per atom, molecular area, area per atom on average, solvent accessible surface, dispersion energy, dielectric energy, H donor, H acceptor, polar and non-polar surface area, atomically resolved H donor, H acceptor, polar and non-polar surface area, shape, sphericity, dipole and higher electric moments, polarizability, dielectric energy, proticity, polar and non-polar surface area, orbital energy and orbital gap, ionization energy, electron affinity, hardness, electronegativity, electrophilicity, excitation energy and intensity, infrared and ultraviolet absorption bands, reactivity measurements, redox potential, bond reference point, partial charge, charge surface area, atomic orbital contribution, bond order, atomic radius. In particular, the physicochemical parameters of the functional compounds preferably refer to 3D descriptors comprising at least one of the following: molecular volume, average volume per atom, molecular area, average area per atom, solvent accessible surface, dispersion energy, dielectric energy, H donor, H acceptor, polar and / or non-polar surface area, atomically resolved H donor, H acceptor, polar and / or non-polar surface area, shape, sphericity, cone angle, polarizability, dielectric energy, proticity, polar and / or non-polar surface area, excitation energy and intensity, infrared and / or UV absorption bands, reactivity measurements, particle charge and / or charge surface area. Preferably, the high-dimensional descriptors utilized may include at least one of the following: configurational partition function, solubility, vapor pressure, activity coefficient, diffusion coefficient, partition coefficient, surface activity, rotational constant, moment of inertia, radius of gyration, density, viscosity, conformer-weighted volume and area, conformer-weighted H donor, H acceptor, proticity, polar and / or non-polar surface area, charge distribution, configurational dipole moment, and molecular refraction. Preferably, high-dimensional descriptors utilized include at least one of solubility, vapor pressure and activity coefficient, surface activity, conformer-weighted H donor, H acceptor, proticity, polar and non-polar surface area, and charge distribution.

[0027] The physicochemical parameters of the functional compounds may also be encoded in molecular graph representations, e.g., molecular fingerprints such as extended connectivity fingerprints, one-hot encoding, word embedding, or graph-level embedding. For example, a machine learning feature model may be utilized to generate predetermined molecular fingerprints containing a predetermined amount of characterization parameters. Such machine learning models may be trained using respective molecular databases in a supervised or unsupervised manner. Preferably, a graph neural network is utilized as the feature model. Preferably, the molecular graph representations reflect similarities in the properties of the functional compounds. In one embodiment, the machine learning-based feature model may be utilized as part of a biodegradation model to determine respective molecular graph representations from the digital representations of the functional compounds, which may then be utilized as input to a second machine learning model as part of the biodegradation model to determine biodegradation based on the molecular graph representations.

[0028] The method further includes providing biodegradation habitats, where the biodegradation habitats indicate habitat descriptor values ​​of habitat descriptors that affect the biodegradation of the functional compound in the respective habitats. In particular, providing may refer to receiving the biodegradation habitats from a user's input, for example, using a respective input unit. Providing may also refer to accessing a storage unit in which the biodegradation habitats are already stored. Providing may also refer to pre-configuring the biodegradable habitats. For example, if the method is used in a very specific situation where only one specific biodegradable habitat is sensible, the respective biodegradable habitats can be pre-configured and do not need to be provided as specific inputs. Providing may also include directly receiving the habitat descriptor values ​​of the habitat descriptors from another source, for example, via a network connection, and providing the received habitat descriptor values ​​of the habitat descriptors as the biodegradation habitats. The provided biodegradation habitat may refer to a general habitat, such as a wastewater habitat, for which the habitat descriptor values ​​of the respective habitat descriptors are already stored in a respective storage that can be accessed. However, the provided biodegradation habitat may also directly include the respective habitat descriptor values ​​of the biodegradation habitat to provide further specification of the biodegradation habitat, such as a marine benthic organism. Furthermore, providing the biodegradation habitat may include providing a digital representation of the biodegradation habitat, where the digital representation may indicate the respective habitat descriptor values ​​of the habitat descriptors that affect the biodegradation of the functional compound in the respective habitat.

[0029] In general, habitat descriptors indicate the environmental characteristics of the habitat. In particular, the environmental characteristics of biodegradation habitats can affect biological activity in each habitat, for example, the presence, growth, or absence of specific microorganisms. Therefore, the environmental characteristics defined by the habitat descriptors also indirectly affect the biodegradation of functional compounds in each habitat. For example, if a functional compound is biodegradable by a specific microorganism that requires a specific salinity, the functional compound will be biodegraded quickly in a habitat that provides such a salinity, such as a marine habitat, but will be biodegraded much more slowly in a habitat that does not provide the appropriate salinity, such as wastewater.

[0030] Preferably, the biodegradation habitat refers to any one of marine habitat, wastewater habitat, lake habitat, anaerobic habitat, compost habitat, or soil habitat. In a preferred embodiment, the biodegradation habitat refers to marine habitat, and the habitat descriptor refers to at least one of salinity, sedimentation type, oxygen level, location, sample depth, water temperature, nutrient concentration, pH value, environment type, and microbial community. In a further preferred embodiment, the biodegradation habitat refers to lake habitat, and the habitat descriptor refers to at least one of salinity, sedimentation type, oxygen level, location, sample depth, water temperature, nutrient concentration, pH value, environment type, and microbial community. In a further preferred embodiment, the biodegradation habitat refers to wastewater, and the habitat descriptor refers to at least one of water temperature, microbial community, sludge concentration, nutrient concentration, pH value, test period, and enzyme environment. Preferably, in the case of a wastewater habitat, the biodegradability model is trained based on biodegradability determined using standard tests determined by the OECD 301 and OECD 302 standards. Furthermore, the habitat descriptor values ​​of this habitat preferably refer to the habitat descriptor values ​​determined by the OECD 301 and OECD 302 standards. In a more preferred embodiment, the biodegradation habitat refers to soil, and the habitat descriptors refer to at least one of temperature, sand content, pH value, moisture content, nutrient concentration, microbial community, and enzyme environment. In a more preferred embodiment, the biodegradation habitat refers to compost, and the habitat descriptors refer to at least one of temperature, compost activity, pH value, moisture content, humidity, compost maturity, compost composition, compost origin, nutrient concentration, microbial community, and enzyme environment. In another preferred embodiment, the biodegradation habitat refers to seawater saltwater, and the habitat descriptor refers to at least one of temperature, salinity, pressure, pH value, nutrient concentration, microbial population, carbonate hardness, calcium concentration, magnesium concentration, nitrate concentration, and phosphate concentration. Generally, the habitat can also refer to the habitat of a standard test used to determine the biodegradability of functional compounds. For example, standard tests such as those specified by ISO13432, ISO14852, ISO14855, ISO17556, OECD 301, and OECD 302 also specify the specific habitat in which biodegradation occurs.Thus, providing a biodegradation habitat can also include providing one of the standard tests, e.g., selecting it via user input, where the habitat descriptor refers to a particular characteristic of the test, i.e., the test environment, and thus the test habitat. Furthermore, a habitat can also be defined by the biodegradation of a reference functional compound or other reference chemical. In this case, the habitat can be provided by providing the reference and its biodegradation. In this case, the reference and its biodegradation represent the habitat descriptor.

[0031] The method further includes providing a biodegradation model based on the provided biodegradation habitat. In particular, providing a biodegradation model preferably refers to selecting a biodegradation model based on the provided biodegradation habitat. For example, multiple biodegradation models may be stored in the biodegradation storage, each trained for a different value or range of habitat descriptor values ​​for one biodegradation habitat, particularly for the biodegradation habitat. Then, based on the provided biodegradation habitat representing the habitat descriptor values, a respective appropriate biodegradation model may be selected from the multiple biodegradation models. For example, a biodegradation model is appropriate if the indicated habitat descriptor value falls within the range of habitat descriptor values ​​for which the biodegradation model was trained. For example, a respective lookup table may be provided that facilitates comparison between the indicated habitat descriptor value and the descriptor value range for which the biodegradation model stored in the storage was trained, so that an appropriate biodegradation model can be directly selected. However, in another embodiment, providing a biodegradation model based on the provided biodegradation habitat may also refer to user selection of a biodegradation model. For example, a user may be provided with a preselection of biodegradation models that refer to the provided biodegradation habitats, and then be able to select the respective biodegradation model to be used. Generally, the possible stored biodegradation models refer to biodegradation models that have already been parameterized based on the respective training data sets of one or more habitats. Because the training data sets used to parameterize the biodegradation models are historical data, as will be explained in more detail below, the biodegradation models may be trained at any time prior to determining the specific biodegradation of a specific functional compound, and thus generated and stored in the respective databases after training. However, training, and therefore generation of the biodegradation model, may of course also be performed when it is determined that a specific biodegradation model for a specific habitat is needed, for example.

[0032] The biodegradation model can be parameterized based on a training dataset that includes measured biodegradation in each habitat associated with each formulation in the training dataset. The measured biodegradation can be measured for each habitat using a predetermined biodegradation test method, such as any of the test methods described above. The biodegradation model thus represents the measured biodegradability of the training formulations.

[0033] The provided biodegradation model is then adapted to determine the biodegradability of the functional compound in each biodegradation habitat. In particular, the biodegradation model is a data-driven model parameterized with respect to the biodegradation habitat, determining the biodegradability of the functional compound based on the digital representation of the functional compound. Preferably, the biodegradation model is trained to determine biodegradation based on the characterization parameters of the functional compound represented by the digital representation. Additionally or alternatively, the biodegradation model may be trained to determine biodegradation based on chemical structure, preferably based on the chemical formulas and amounts or ratios of two or more structural formulas present in the functional compound, as described above. The term "to" should be interpreted herein as meaning that the parameterization adapts the biodegradation model when the physicochemical parameters of the functional compound are provided as input, thereby enabling the biodegradation model to provide biodegradability for the habitat. For example, the biodegradation model relates the physicochemical parameters of the functional compound of the previous digital representation of the functional compound and the previous digital representation of the habitat to biodegradability. This allows the digital representation of the functional compound to be determined based on the target biodegradability. The term "data-driven" is used herein to emphasize that the model is primarily based on the respective data inputs and not, for example, on intuition, personal experience, or knowledge. Preferably, the biodegradation model refers to a machine learning-based model based on known machine learning algorithms, such as neural networks, regression models, classification algorithms, etc. In this context, regression models based on neural networks, linear regression, random forests, boosted trees, Lasso, ridge regression, and MARS algorithms have been found to be particularly suitable for most applications, while random forests, logistic regression, and SVM algorithms have been found to be particularly suitable for classification models. Generally, the biodegradation model is parameterized during the training process, where a digital representation of a functional compound or one or more characterization parameters and / or chemical structure, as described above, is utilized along with the corresponding biodegradability in a specific biodegradation habitat.Based on such a training data set specific to a biodegradation habitat, e.g., a particular habitat descriptor range and / or value, known training methods can be used to determine the respective parameters of the data-driven model so that the biodegradation model can also determine the biodegradation of functional compounds that are not part of the training data set.

[0034] In one embodiment, the biodegradation model is a two-stage machine learning model including two machine learning algorithms, the output of the first stage being the input of the second stage, and both machine learning algorithms are trained simultaneously using the same training dataset. Preferably, the first stage is trained to determine one or more characterization parameters from the digital representation, and then the second stage is trained to use the one or more characterization parameters to determine biodegradability. In a preferred embodiment, the first stage is based on a graph neural network trained to determine a molecular graph representation, such as a molecular fingerprint, of the functional compound based on the digital representation, and the second stage is based on a random forest algorithm to determine the biodegradation of each based on the molecular graph representation and optionally habitat descriptors.

[0035] Furthermore, in a preferred embodiment, the biodegradation model can also be adapted to determine the biodegradation of the functional compound further based on habitat descriptor values ​​as input. In particular, the biodegradation model can be trained by utilizing a training dataset that includes a) a digital representation of the functional compound, preferably including or representing, for example, the functional compound's physicochemical parameters, structural formula, etc., and b) the associated biodegradability in a specific habitat, thereby generating a biodegradation model that indirectly takes that specific habitat into account. However, the training dataset may optionally also include specific habitat descriptor values ​​for each habitat. In this case, in addition to the digital representation, the habitat descriptor values ​​may also be provided as input to the biodegradation model, and the biodegradation model can be trained to determine biodegradability further based on the habitat descriptor values. This has the advantage that biodegradation can be determined even more accurately, especially when biodegradation strongly depends on the specific habitat descriptor values ​​of the habitat. For example, in marine habitats, temperature or salinity can vary greatly in different regions of the world, which can lead to different biodegradability for some functional compounds. Therefore, for such cases, it can be advantageous to provide habitat descriptor values ​​directly as inputs to the biodegradation model. However, instead of providing habitat descriptor values ​​as inputs to the biodegradation model, it is also possible to train two different biodegradation models and indirectly treat different regions as different habitats.

[0036] The method further includes determining the biodegradability of the functional compound based on the provided biodegradation model and the digital representation of the functional compound. In particular, as described above, the amounts of two or more structural formulae present in the functional compound and / or characterization parameters provided by or derivable from the digital representation of the functional compound can be provided as inputs to the biodegradation model. The biodegradation model then provides the determined biodegradability as an output. If the digital representation does not directly include the respective inputs, e.g., physicochemical parameters, determining the biodegradability can also include first determining the respective inputs, e.g., the structural formulae present in the functional compound, the amounts of two or more structural formulae, and / or characterization parameters. Such determined inputs can then be provided to the biodegradation model. The determined biodegradability can then be provided, for example, to an output unit or a computing unit for further processing. Preferably, providing the biodegradability leads to further processing that utilizes the determined biodegradability. In such cases, providing as an individual step can be omitted and replaced by processing the determined biodegradability.

[0037] The determination of biodegradability using a biodegradation model can be considered a virtual measurement of biodegradability. In particular, the biodegradation model is based on measurement data, e.g., the measured biodegradability of the functional compound used to train the biodegradation model. Thus, the biodegradation model includes information provided by these previous measurements. Furthermore, inputs to the biodegradation model, e.g., physicochemical parameters or structural formulas, may also refer to measured properties of the functional compound in some cases. Thus, the biodegradability of a new functional compound determined using a biodegradation model can also be considered to be based at least in part on measurement results. Preferably, providing the target functional compound leads to further processing to determine potential synthesis specifications for the target functional compound. This can be done, for example, by querying a database with stored synthesis specifications or using a data-driven forward synthesis planning tool.

[0038] Preferably, the biodegradability processing includes determining a control signal for controlling a manufacturing process based on the determined estimated biodegradability. The manufacturing process may refer to the manufacturing process of the functional compound or the biodegradation process of a product in which the functional compound is utilized. For example, for controlling a biodegradation process, if the determined biodegradability indicates that the functional compound will biodegrade at an appropriate rate in a particular environment, i.e., habitat, generating a control signal may include generating a control signal for controlling a waste management facility to provide this habitat, for example, by providing a respective temperature. In a preferred embodiment of the manufacturing process of the functional compound, the control signal indicates a machine-executable synthesis specification for the functional compound, particularly if the comparison indicates that the determined biodegradability of the functional compound is within a predetermined range around the provided target biodegradability. For example, the synthesis specification for the functional compound may be known, for example, stored in a respective database, or may be determined using known methods for determining synthesis specifications for compounds.

[0039] Furthermore, processing biodegradability may also refer to selecting one or more functional compounds based on their respective determined biodegradability. For example, if the respective biodegradability of multiple potential functional compounds has been determined, selecting may include comparing the biodegradability of different functional compounds to predetermined selection criteria and selecting functional compounds whose determined biodegradability meets these criteria. In particular, in one embodiment, the method includes receiving a target biodegradability of a functional compound, comparing the received target biodegradability with the determined biodegradability, and providing a control signal in response to the comparison. The control signal may refer to any signal that enables further control of a technical system. For example, the control signal may be adapted to control an interface to provide the comparison result on an interface. In a preferred embodiment, the comparison refers to verification of the target biodegradability, and the verification is positive if the determined biodegradability is within a predetermined range around the target biodegradability. In this case, the control signal may be adapted to simply control a user interface to provide an indication of whether the verification result is positive or negative. Preferably, however, the control signal refers to a recipe, i.e., a synthesis specification, of one or more functional compounds that meet the specified target biodegradability, i.e., are verified as positive. A recipe, i.e., a synthesis specification, is generally defined as an instruction on how to synthesize a functional compound. Specifically, a recipe includes starting materials and respective parameters for polymerization from the starting materials. Such a recipe may already be known for the functional compound, e.g., stored in a respective database, or may be determined using known theory or experimental methods, e.g., using a respective known model for determining synthesis specifications from input compounds. Preferably, the control signal includes the recipe in a form that directly enables automatic control of a respective industrial system or industrial equipment for producing the functional compound. In particular, if the result of the comparison indicates that the determined biodegradability is within a predetermined range around the target biodegradability, it is preferred that the control signal indicates a machine-executable synthesis specification for the functional compound.

[0040] In a preferred embodiment, the method further includes providing connectivity information as a digital representation of the functional compound and determining model inputs, e.g., structural formula, characterizing parameters, from the connectivity information. In particular, the connectivity information includes information about the atoms, chemical bonds between them, and stereochemistry of the functional compound. The method then includes determining the chemical structure, structural formula, and / or characterizing parameters from the connectivity information.

[0041] In one embodiment, the method further includes providing a biodegradation test method, where the provided biodegradation test method represents a standardized biodegradation test method for experimentally determining the biodegradation of a chemical substance, and a biodegradation model is further provided based on the provided biodegradation test method. Generally, multiple standardized biodegradation test methods exist for testing the biodegradability of chemical substances. For example, such test methods may be found in DIN or ISO standards. Furthermore, by providing a biodegradation test method and a biodegradation model trained using the provided biodegradation test method, each test method can be used to determine biodegradation that can be easily compared to, for example, each measured biodegradation. In particular, in this embodiment, the biodegradation model is preferably trained based on a dataset, and the test methods by which biodegradation was determined are clearly specified so that the biodegradation model can be specifically trained for one or more test methods.

[0042] In one embodiment, a target use of the functional compound is further provided, which indicates the intended use of the target functional compound, and a biodegradation habitat is provided based on the target use. The target use of the functional compound may, for example, indicate the intended use situation of the functional compound, such as whether the functional compound is intended to be used as an excipient, plasticizer, stabilizer, inhibitor, odorant, fragrance component, nutrient component, catalyst, radiation absorber, lubricant, or surfactant. Such a target use indicates a specific biodegradation habitat. For example, in the case of a pesticide used in agriculture, it may be interesting to know whether the functional compound will biodegrade in soil. In another example, if the target use indicates the use of the functional compound as a surfactant in personal care products, the functional compound will most likely be found in wastewater environments sooner or later. In this manner, each target use may indicate a respective biodegradation habitat. In this regard, a predetermined list may be provided in a storage device in which each target use and its corresponding biodegradation habitat are stored. The target use of the functional compound can then be provided, for example, by providing the user with a list of target uses and allowing the user to select an individual target use, with each individual target use being connected to one or more biodegradation habitats. Biodegradability can then be determined for each of the biodegradation habitats to which the target use is connected, or again, the user can select an individual biodegradation habitat to which the target use is connected. Additionally or alternatively, information indicating the intended end-of-life treatment of the functional compound can be provided. For example, the end-of-life treatment can indicate whether the functional compound is intended to biodegrade in a particular environment or whether it should be subjected to a particular treatment, for example, in a bioreactor. Thus, the intended end-of-life treatment information can also be used to determine the biodegradation habitat of the functional compound, as described above.

[0043] In one embodiment, further information indicating the accessible surface area of ​​the functional compound in its intended form is provided, and the biodegradation model is further trained to determine biodegradability based on the accessible surface area, and the method further comprises further determining biodegradability based on the accessible surface area. For example, the information may indicate whether the intended product is provided in a solid, crushed, foamed, pelleted, or any other form. Preferably, the information indicates the surface area of ​​the product per mass or the geometric shape of the smallest independent part of the product. Generally, the biodegradability of a functional compound is an intrinsic property of the functional compound, but the precise timing of biodegradation of a product containing the functional compound may also depend, for example, on the surface area accessible by microbial components of the habitat involved in biodegradation. Therefore, further determining biodegradability based on the surface area of ​​a product containing the functional compound can increase the accuracy of predicting the biodegradability of the final product.

[0044] In one embodiment, habitat descriptor values ​​for the habitat descriptors are stored in association with respective geographic locations, and providing a biodegradation habitat refers to providing the geographic location of the habitat and retrieving the habitat descriptor value for that geographic location from storage. The geographic location may refer to, for example, coordinates or other area identifiers. For example, the geolocation may refer to a city name, a country name, a country region name, a sea area name, a geographic feature, etc. Based on such geolocation, the average or minimum and maximum values ​​of each habitat descriptor, for example, the habitat descriptor, may be stored. Thus, by providing a geolocation, the respective habitat descriptor value for that geolocation may be provided. This has the advantage that the user does not need to know the exact habitat or the exact habitat descriptor value in a certain area. Thus, the user can simply provide a location where a functional compound is expected to biodegrade in that area.

[0045] In one embodiment, the characterization parameters indicated by the digital representation of the functional compound may refer to at least one of a compositional descriptor, a count descriptor, a list of structural fragments, a fingerprint, a graph invariant, a 3D descriptor, and / or a higher-dimensional descriptor that indicate the chemical properties of the functional compound. The respective connections between the digital representation and, for example, previously calculated functional compound descriptors or further information about the functional compound may already be stored and connected to the respective digital representation. For example, if the digital representation refers to a brand name or other identifier, the respective structural formula, the ratio of the structural formula in a particular habitat, and / or physicochemical parameters corresponding to the brand name or identifier may already be stored, for example, in the brand name owner's storage.

[0046] In one embodiment, the method further includes providing a biodegradation test method, wherein the provided biodegradation test method represents a standardized biodegradation test method for experimentally determining biodegradation of a chemical substance, and a biodegradation model is further provided based on the provided biodegradation test method. For example, the test method may be any of the following standardized test methods: ISO 13432, ISO 14852, ISO 14855, ISO 17556, OECD 301, and OECD 302.

[0047] In a further aspect, an interface method for providing an interface is presented, the interface method including: i) receiving as input a digital representation and a habitat via a user interface, and providing the received digital representation and habitat to a processor executing the method; and ii) providing as a result, via the user interface, a determined biodegradability of the functional compound to a user, the result being received from the processor executing the method.

[0048] In a further aspect, a computer-implemented training method for training a data-driven biodegradation model to parameterize the biodegradation model is presented, the training method including: i) providing training data related to a predetermined biodegradation habitat, the training data including: a) digital representations of a plurality of training functional compounds; and b) biodegradability of each biodegradation habitat associated with each training functional compound; ii) providing a data-driven trainable biodegradation model; iii) training the provided data-driven biodegradation model based on the provided training data, the trained biodegradation model being adapted to determine the biodegradation of the functional compounds based on the digital representations of the functional compounds; and iv) providing the trained biodegradation model.

[0049] In a further aspect, an apparatus for determining the biodegradability of a predetermined functional compound is provided, the apparatus comprising: i) a digital representation providing unit for providing a digital representation of the functional compound; ii) a habitat providing unit for providing biodegradation habitats, the biodegradation habitats indicating habitat descriptor values ​​of habitat descriptors that affect the biodegradation of the functional compound in each habitat, the habitat descriptors indicating environmental characteristics of the habitat; iii) a model providing unit for providing a biodegradation model based on the provided biodegradation habitats, the biodegradation model being adapted to determine the biodegradability of the functional compound in each biodegradation habitat, the biodegradation model being a data-driven model parameterized with respect to the biodegradation habitat to determine the biodegradability of the functional compound based on physicochemical properties; and iv) a determining unit for determining the biodegradability of the functional compound based on the selected biodegradation model and the digital representation of the functional compound.

[0050] In a further aspect, an interface device for providing an interface is presented, the interface device comprising: i) an input interface unit for receiving, as input, a digital representation and a habitat via a user interface and providing the received digital representation and habitat to said device; and ii) a result interface for providing, as a result, the determined biodegradability of the functional compound to a user via the user interface, the result being received from said device.

[0051] In a further aspect, a training device for training a data-driven biodegradation model for parameterizing a biodegradation model is presented, the training device including: i) a training data providing unit for providing training data related to a predetermined biodegradation habitat, the training data including: a) digital representations of a plurality of training functional compounds; and b) biodegradability of each biodegradation habitat associated with each training functional compound; ii) a trainable model providing unit for providing a data-driven based trainable biodegradation model; iii) a training unit for training the provided data-driven biodegradation model based on the provided training data, the trained biodegradation model being adapted to determine the biodegradation of the functional compounds based on the digital representations of the functional compounds; and iii) a trained model providing unit for providing a trained biodegradation model.

[0052] In a further aspect of the present invention, the use of the above-mentioned method is provided, wherein the method is used to determine the biodegradability of a given functional compound in any of the following: i) a functional compound referred to as a nutritional ingredient, ii) a functional compound referred to as a UV absorber used in skin protection, iii) a functional compound referred to as a formulation additive used in personal care applications, iv) a functional compound used in aroma applications, v) a functional compound used as a plasticizer, vi) a functional compound used as a lubricant, and vii) a functional compound used as an active ingredient.

[0053] In a further aspect of the present invention, a system is provided that includes: i) a control signal that includes a synthesis specification for a functional compound, the synthesis specification indicating one or more ingredients for producing the functional compound, the control signal being generated according to the method described above; and ii) one or more ingredients indicated by the synthesis specification in the control signal.

[0054] In a further aspect of the present invention, there is provided the use of a control signal generated according to the above-described method for controlling a manufacturing process, particularly a manufacturing process comprising the production of a functional compound. In a further aspect of the present invention, there is provided a control signal, the control signal generated according to the above-described method. Preferably, the control signal comprises a machine-executable synthesis specification for producing a target functional compound.

[0055] In a further aspect, a computer program product for determining the biodegradability of a given functional compound is provided, the computer program product comprising program code means for causing the above-described apparatus to perform the above-described method.

[0056] In a further aspect, a computer program product for training a biodegradation model is presented, the computer program product comprising program code means for causing the training device to perform the training method described above.

[0057] It is to be understood that the above-mentioned method, the above-mentioned device and the above-mentioned computer program product have similar and / or identical preferred embodiments, in particular as defined in the dependent claims. Furthermore, the above-mentioned training method, the above-mentioned training device and the above-mentioned training computer program product also have similar and / or identical preferred embodiments, in particular as defined in the dependent claims.

[0058] It is to be understood that a preferred embodiment of the invention can also be any combination of the dependent claims or the above embodiments with the respective independent claim.

[0059] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter. [Brief explanation of the drawings]

[0060] [Figure 1] 1 illustrates, in a schematic and exemplary manner, one embodiment of a system including an apparatus for determining the biodegradability of a functional compound. [Figure 2] 1 shows, by way of example, a flow chart of a method for determining the biodegradability of a functional compound. [Figure 3] 1 shows, by way of example only, a flow chart of a method for training a biodegradation model for determining the biodegradability of a functional compound. [Figure 4] 1 shows, by way of example only, a flow chart of one embodiment of a method for determining the biodegradability of a functional compound. [Figure 5] 1 shows, schematically and exemplarily, a block diagram of the system architecture of a system and apparatus for determining the biodegradability of functional compounds. [Figure 6] 1 shows, schematically and exemplarily, a block diagram of the system architecture of a system and apparatus for determining the biodegradability of functional compounds. [Figure 7] 1 shows, schematically and exemplarily, a block diagram of the system architecture of a system and apparatus for determining the biodegradability of functional compounds. DETAILED DESCRIPTION OF THE INVENTION

[0061] Detailed Description of the Embodiments 1 shows a schematic and exemplary embodiment of a system 100 including an apparatus 110 for determining the biodegradability of a functional compound based on a digital representation of the functional compound and a provided biodegradation habitat. The system 100 further includes a training apparatus 130 for training a biodegradation model utilized in the apparatus 110, a database 140 in which the results of the determination of the biodegradability of the functional compound can be stored, and a manufacturing system 120 for manufacturing products, particularly products including the functional compound, that can be controlled using the determined biodegradability.

[0062] The apparatus 110 includes a digital representation providing unit 111, a habitat providing unit 112, a model providing unit 113, a determining unit 114, and optionally an output and / or control unit 115 that may be adapted to output the determined biodegradability and / or provide a control signal for controlling a manufacturing process of the manufacturing system 120 based on the determined biodegradability.

[0063] The digital representation providing unit 111 is adapted to provide a digital representation of the functional compound. The digital representation providing unit 111 may, for example, refer to an input unit through which a user can input the respective digital representation. Furthermore, the digital representation providing unit 111 may refer to or be part of a user interface that allows a user to interact with the device 110 and / or the database 140. However, the digital representation providing unit 111 may also refer to or be communicatively coupled to a storage unit in which the digital representation of the functional compound is already stored. Generally, the digital representation may directly include the chemical structure, the structural formula, and the respective amounts of two or more structural formulas present in the functional compound, and / or characterizing parameters of the functional compound. However, for example, instead of directly providing the characterizing parameters, it is also possible to provide only the known synthesis specifications and / or molecular structure of the functional compound. In this case, it is preferable that the digital representation providing unit 112 is further adapted to determine respective other information and / or parameters, for example, the chemical structure, the structural formula, and the respective amounts of two or more structural formulas present in the functional compound, and / or characterizing parameters from the synthesis specifications and / or molecular structures. In particular, the digital representation providing unit 111 may be adapted to determine the chemical structure, the structural formula and the amount of each of two or more structural formulas present in the functional compound and / or to determine characterizing parameters, for example by accessing a database in which respective information for a plurality of functional compounds is stored, and then the digital representation providing unit 111 is adapted to provide a digital representation, for example to the determining unit 114, each comprising the determined further information, such as physicochemical parameters.

[0064] The habitat providing unit 112 is adapted to provide biodegradation habitats. The habitat providing unit 112 may, for example, refer to an input unit through which a user may input the respective biodegradation habitat. For example, a user interface may be provided that allows the user to select from several predetermined biodegradation habitats. In a preferred embodiment, the habitat providing unit may be communicatively coupled to or refer to a user interface that allows the user to indicate a geolocation, for example, by marking a location on a map, indicating coordinates, or providing a name of an area, such as a political or geological area. The habitat providing unit may then be adapted to provide a biodegradation habitat based on the geolocation. For example, if the geolocation indicates a particular sea area, such as the North Sea or the Atlantic Ocean, the habitat providing unit may be adapted to determine a marine habitat as the biodegradation habitat.

[0065] Generally, biodegradation habitats indicate habitat descriptor values ​​of habitat descriptors that affect the biodegradation of functional compounds in each habitat. In particular, habitat descriptors indicate the environmental characteristics of the habitat; for example, in the case of marine habitats, salinity can strongly affect the biodegradation of functional compounds in marine habitats. Generally, the chemical effect of the habitat descriptor on the functional compound is not important in this application because biodegradation is determined. Therefore, it is the effect of the habitat descriptor on the habitat ecology, particularly the microbial population of the habitat, that indirectly affects biodegradation. Typical specific habitat descriptor values ​​for each habitat can be stored in the database. However, for example, if it is known that the habitat descriptor value for each habitat deviates from the typical habitat descriptor value, the user can also input each specific habitat descriptor value.

[0066] The model providing unit 113 is adapted to provide a biodegradation model based on the provided biodegradation habitat. In particular, the model providing unit 113 is preferably adapted to select a biodegradation model from a plurality of biodegradation models already stored in the database. For example, the biodegradation model may be trained on training data corresponding to one or more specific biodegradation habitats. These specific biodegradation habitats may be defined in terms of habitat descriptor values ​​or ranges that define for which biodegradation habitats the respective biodegradation model is suitable. For example, a look-up table may be provided that allows the model providing unit to select which biodegradation model is suitable based on the biodegradation habitat, e.g., based on the habitat descriptor values ​​of the biodegradation habitat. However, the model providing unit 113 may also include or point to an input unit that can receive a biodegradation model, e.g., by user selection or user input indicating which biodegradation model should be used.

[0067] The biodegradation model is a parameterized data-driven model for determining the biodegradability of a functional compound based on a digital representation, particularly based on the chemical structure, structural formula, and amount of two or more structural formulas present in the functional compound and / or characterization parameters associated with the functional compound. Optionally, the biodegradation model may also be trained to further utilize habitat descriptor values ​​provided as input. In a preferred embodiment, the data-driven model refers to a machine learning model that utilizes, for example, a regression model-based algorithm or a classification model-based algorithm. The regression model-based algorithm may be based on any of a neural network algorithm, a linear regression algorithm, a LASSO algorithm, a ridge regression algorithm, a MARS algorithm, a random forest algorithm, and a boosted tree algorithm. The classifier model algorithm may be based on any of a random forest algorithm, a logistic regression algorithm, and an SVM algorithm. The inventors have found that for most applications, neural network, linear regression, random forest, and MARS-based algorithms are particularly suitable.

[0068] The biodegradation model can be trained, for example, using a training device 130. In particular, the training device 130 includes a training data providing unit 131 for providing training data for training the data-driven biodegradation model. The training data includes a) digital representations of a plurality of training functional compounds for one or more different habitats, and b) a biodegradability associated with each training functional compound. Optionally, the training dataset may further include habitat descriptor values ​​for the specific habitats in which the biodegradability of each of the functional compounds has been determined. Preferably, in the training data, the biodegradability provided for each training functional compound refers to biodegradability measured according to the same measurement method. However, biodegradability may also be provided for different measurement methods, in which case it is preferably clearly indicated which biodegradability is associated with which measurement method so that the biodegradation model can be trained to distinguish between the different measurement methods. Generally, the training data can be designed to cover a predetermined habitat space of the biodegradation model to be trained, where the habitat space is defined by the range of values ​​of each habitat descriptor for which the biodegradation model is trained. For example, training data can be designed to cover a given functional compound type for a given habitat. Known methods for designing and optimizing training data for a given habitat space can be used to ensure that the habitat space is adequately covered by the training data and that random outliers are avoided.

[0069] Further, the training device 130 includes a model providing unit 132 adapted to provide a data-driven, trainable biodegradation model, e.g., a biodegradation model including parameters that can be set during a training process to train the biodegradation model. For example, the trainable biodegradation model may be already stored in a storage unit that the model providing unit 132 can access to provide it. Further, the training device 130 includes a training unit 133 for training the provided data-driven biodegradation model based on the provided training data. In particular, training may refer to changing parameters of the biodegradation model based on the respective training data until the biodegradation model is adapted to determine the biodegradability of the functionalized compound based on the digital representation. Generally, any known training algorithm for training a data-driven, particularly machine learning-based, model may be utilized. Preferably, during training of the biodegradation model, input parameters, e.g., any of the chemical structure, structural formula, and amount of two or more structural formulas present in the functionalized compound and / or characterization parameters of the functionalized compound that most affect biodegradation in each habitat, are also determined, and the model is then trained based on these most influential input parameters. To determine these most influential input parameters, for example, cluster analysis or PCA analysis tools can be used. Furthermore, learned representations based on molecular graphs can be used as input for predictive models. In particular, the input parameters can be used to determine the application space of the training data, where the application space is defined by the physicochemical parameters of the functional compounds and the habitat descriptors covered by the data. The determination of the most influential physicochemical parameters and / or habitat descriptors can then be performed as a dimensionality reduction of the application space. An algorithm can then be applied to optimize the training data within the application space, for example, to cover the application space with as little training data as possible.

[0070] The training device 130 then comprises a trained model providing unit 134 adapted to provide the trained biodegradation model to a storage unit in which trained biodegradation models for different habitats and / or different types of functional compounds, respectively, are stored. However, the trained model providing unit 134 may also be adapted to provide the trained biodegradation model directly to, for example, the biodegradation model providing unit 113 of the device 110.

[0071] In all cases, the biodegradation model providing unit 113 is then adapted to provide a suitable trained biodegradation model to the characterization determining unit 114. The determining unit 114 may then utilize the biodegradation model and the provided digital representation to determine the biodegradability. In particular, the determining unit 114 may be adapted to utilize any of the input parameters indicated by the digital representation as input to the trained biodegradation model, such as the chemical structure, the structural formula and the amounts and / or characterization parameters of two or more structural formulas present in the functional compound, as already explained above, to provide as output a decision on the trained biodegradability.

[0072] An output unit, e.g., a display, may then be adapted to output the determined biodegradability. However, the output unit may additionally or alternatively be adapted to provide the determined biodegradability to a database 140 for storing functional compounds in association with the determined biodegradability for future use. In particular, the output unit may be adapted to select each functional compound based on predetermined criteria regarding biodegradability, if the biodegradability for different functional compounds has already been determined and stored, e.g., in the storage unit, i.e., database 140. The output unit may then be adapted to provide and / or output the selected functional compound and its biodegradability. This is particularly suitable when a user searches for a functional compound with a specific biodegradability in one or more habitats from a plurality of candidate functional compounds.

[0073] Optionally, the apparatus 110 may include a control unit 115 adapted to provide a control signal for controlling a manufacturing process of the manufacturing system 120 based on the determined biodegradability. In particular, the control unit 115 is adapted to receive a target biodegradability of the functional compound, compare the received target biodegradability with the determined biodegradability, and provide a control signal in response to the comparison, preferably a control signal indicating the use or production of the functional compound whose biodegradability has been determined. Furthermore, the control signal may indicate a machine-executable synthesis specification for the functional compound whose biodegradability has been determined if the result of the comparison indicates that the determined biodegradability is within a predetermined range around the target biodegradability. However, the control unit 115 may also be adapted to control a manufacturing process of another product based on the determined biodegradability, for example, by providing a control signal indicating a machine-executable synthesis specification for another product that utilizes or includes the respective functional compound. Furthermore, the control unit 115 may provide a control signal for controlling a habitat for biodegrading the functional compound, for example, in a waste treatment facility. For example, target biodegradability may be met for certain habitat descriptors, and the control unit 115 may be adapted to provide control signals that control the facility such that these habitat descriptor values ​​are met.

[0074] FIG. 2 schematically and exemplarily illustrates a flowchart of a method for determining the biodegradability of a functional compound. The method 200 includes a first step 210 of providing a digital representation of the functional compound. Specifically, providing the digital representation in this step may follow the principles described above with respect to the digital representation providing unit 111. Furthermore, in step 220, biodegradation habitats are provided, each indicating habitat descriptor values ​​of habitat descriptors that affect the biodegradation of the functional compound in the respective habitat. For example, the principles described above with respect to the habitat providing unit 112 may also be applied to this step 220. Furthermore, in step 230, a biodegradation model is provided, adapted to determine the biodegradability of the functional compound based on the digital representation. As already described in more detail above, providing a biodegradation model may also refer to the selection of a biodegradation model based on the provided biodegradation habitat. Furthermore, the biodegradation model is a data-driven model parameterized with respect to the biodegradation habitat, such that the biodegradability of the functional compound can be determined based on the digital representation. Generally, steps 210, 220, and 230 can be performed in any order or even simultaneously. In a next step 240, the biodegradability is determined based on the provided digital representation of the functional compound and the biodegradation model. In an optional step 250, the biodegradability can then be provided to a user interface, e.g., so that the determined biodegradability of the functional compound is displayed on a display. However, in step 250, the method can additionally or alternatively include generating a control signal that allows control of a manufacturing process of a product, e.g., the functional compound or a product comprising the functional compound, as already described in detail above.

[0075] FIG. 3 schematically and exemplarily illustrates a flowchart of a method for training a data-driven biodegradation model, for example, as utilized in the method 200 described with reference to FIG. 2. Generally, the method 300 may be performed by, for example, each unit of the training device 130 described with reference to FIG. 1. The method 300 includes a step 310 of providing training data for training the data-driven biodegradation model. The training data includes a) digital representations of a plurality of training functional compounds and b) biodegradability associated with each training functional compound in each biodegradation habitat, for example, for a particular habitat descriptor value. In particular, the training data may be provided according to the principles described above with reference to the training data providing unit 131 described with reference to FIG. 1. The method further includes a step 320 of providing a data-driven trainable biodegradation model, for example, a machine learning-based biodegradation model such as a neural network. Generally, steps 310 and 320 may be performed in any order or even simultaneously. The method 300 then further includes step 330 of training the provided data-driven biodegradation model based on the provided training data, e.g., by varying parameters in the data-driven trainable biodegradation model, such that the trained biodegradation model is adapted to determine the biodegradability of the functionalized compound based on the digital representation of the functionalized compound. In step 340, the trained biodegradation model may then be provided, e.g., by storing the trained biodegradation model in storage or by providing the trained biodegradation model directly to the device 130, as described with respect to FIG. 1.

[0076] A more detailed preferred example of the above-mentioned method and corresponding apparatus will be described below. An exemplary embodiment of the method may consist of the following steps. FIG. 4 provides a schematic and exemplary flowchart of an exemplary embodiment of the method. In this exemplary embodiment, the method begins with requesting a digital representation of a new functional compound, for example, via a user interface. Furthermore, a target application of the new functional compound may also be requested, for example, via the user interface. The target application may refer, for example, to how the new functional compound, for which a digital representation is provided, should be utilized in a product or which waste treatment is expected for the functional compound. Furthermore, in addition to or instead of the target application, a biodegradation habitat may also be requested. However, in that case, each target application may also, for example, indicate a respective biodegradation habitat. For example, a list may be presented to the user via the user interface, from which the user may select each target application. Based on each selection, a selection of biodegradation habitats associated with the target application may also be provided for the user to select. Based on the target application, particularly based on the biodegradation habitat indicated by the target application, a respective biodegradation model, i.e., a biodegradation model, may be selected. Optionally, further preselected conditions indicated by the selected biodegradation model may be required. For example, the biodegradation model may be adapted to utilize further descriptors, such as optional habitat descriptors, application constraints, etc., which may be required as needed and may enable the biodegradation model to determine biodegradation with further accuracy or specifically for the constraints. Furthermore, characterization parameters of, for example, functional compounds may be derived from the provided digital representation input parameters, and the input parameters utilized may also depend on the provided target application. By utilizing the selected biodegradation model and the derived physicochemical parameter values, a respective determined biodegradability, i.e., a respective biodegradability, may be provided.

[0077] FIG. 5 shows a block diagram of an exemplary system architecture of an automated laboratory system 1000 for synthesizing functional compounds, including a laboratory equipment control device 1102, a network 1150, and a synthesis specification, or recipe, module 1100 / 1110, and a client device 1108. The automated laboratory system includes a laboratory equipment control device layer 1152 as part of the laboratory equipment control device 1102, a synthesis specification module layer 1154 associated with the synthesis specification module, and a remote control or client layer 1156 associated with the client device 1108. The laboratory equipment control device layer can be divided into several hierarchical layers: a hardware layer, a middleware layer, and an interface layer. The hardware layer specifically relates to hardware resources, such as sensors and actuators, for controlling the synthesis of functional compounds. The middleware relates to any of the known middleware for laboratory or plant synthesis operations. One example is LABS / QM, which provides various abstractions for hardware, networks, and operating systems, such as low-level device control and message passing. The communication layer relates to communication protocols, one of which is REST, which can be implemented over different transport protocols (i.e., UDP, TCP, telemetry) and allows the exchange of messages between the laboratory equipment control device and the laboratory equipment device. Such a software architecture makes it possible to control and monitor laboratory equipment without interacting with the hardware.

[0078] The synthesis specification module layer 1154 may include a mass storage layer, a computing layer, and an interface layer. The storage layer is configured to provide mass storage for a data-driven biodegradation model to provide recipes, i.e., synthesis specifications, for biodegradable functional compounds, as described in detail above. In particular, the functions performed by the device as described above may be provided as program code stored in the mass storage. Furthermore, synthesis specifications for multiple functional compounds may be stored in the mass storage. Such data may be stored in a structured database, such as an SQL database, a distributed file system, such as HDFS, or a NoSQL database, such as HBase or MongoDB. The computing layer may include an application layer that customizes functionality provided by standard cloud services to perform computing processes based on target properties. Such functions may include determining a digital representation of a target functional compound based on the target biodegradability and biodegradation model, generating a synthesis specification from the digital representation of the target functional compound, and providing the synthesis specification as control data to a laboratory equipment control device.

[0079] The interface layer may implement a web service, a network interface as UDP or TCP, or a web socket interface. For communication with laboratory equipment control devices, a REST API is implemented.

[0080] The client layer 1156 provides an interface to an end user. To the end user, the client layer 1156 can execute a client-side web application that provides an interface to the composite specification module layer 1154 or the laboratory equipment control device layer 1152. The user may be provided with a UI for selecting a target biodegradability and a biodegradation habitat for the target biodegradability, which may also include a range of biodegradability values. In another example, the user may be provided with a UI for selecting multiple target biodegradabilities and their respective values. The application may be configured for a user to remotely monitor and control the laboratory equipment control device and operation. In another example, the client device layer and composite specification module layer may be incorporated into a single device. The alternatives described herein are merely illustrative and should not be construed as limiting.

[0081] 6 shows a block diagram of an exemplary system architecture of a system and apparatus for generating a biodegradation model for determining biodegradability, namely, a model generation module 2100 / 2110, which may be considered as or include a network 2150 and a training model apparatus, a synthesis specification module 1100 / 1110, and a client device 2108. The system for generating a biodegradation model includes a model generation module layer 2154 as part of the model generation module and a client layer 2156 associated with the client device 2108.

[0082] The model generation module layer 2154 may include a mass storage layer, a computing layer, and an interface layer. The storage layer is configured to provide mass storage for the data-driven biodegradation model described above. Furthermore, the mass storage is configured to store the synthesis specifications of the functional compounds and the measured biodegradability of one or more habitats. Such data may be stored in a structured database such as an SQL database, a distributed file system such as HDFS, or a NoSQL database such as HBase or MongoDB. The computing layer may include an application layer that enables customization of functions provided by standard cloud services to execute computing processes for generating a biodegradation model for determining the biodegradability of the functional compounds. Such functions may include receiving, in a model generation module, digital representations of at least two previously measured functional compounds associated with a synthetic specification, and at least one measurement data of biodegradability in at least one habitat for each of the at least two previously measured functional compounds; receiving a digital representation of at least one unmeasured functional compound; training a model according to the training principles described above based on a similarity measure between the digital representations of the at least two previously measured functional compounds, the measurement data of biodegradability in at least one habitat for each of the at least two previously measured functional compounds, and preferably the digital representations associated with the synthetic specification for each of the at least two previously measured functional compounds and the digital representations associated with the synthetic specification for each of the at least one unmeasured functional compound; and providing a biodegradation model via an output interface. The model generation module layer may be configured to deploy the generated model and synthetic specification database to the synthetic specification module layer. This may include storing the generated model and synthetic specification database in a mass storage device associated with the synthetic specification module.

[0083] The model generation module layer may be further configured to determine a digital representation of the functional compound associated with the synthesis specification and / or molecular structure. The digital representation may include a set of physicochemical parameters and functional compound parameter values ​​associated with the synthesis specification and / or molecular structure of each measured functional compound. When a model is generated based on a digital representation derived from a recipe, the relationship between the synthesis specification and / or molecular structure and the physicochemical parameters may be stored in a mass storage device associated with the model generation module. In such a case, deploying the model includes providing the relationship.

[0084] The interface layer may implement a web service, a network interface as UDP or TCP, or a web socket interface. In this example, a REST API is implemented for communication with client devices. The client layer 2156 provides access to a mass storage device containing synthetic specifications for functional compounds and at least one biodegradability for at least two functional compounds. The client layer also provides an interface to an end user. For end users, the client layer 2156 can execute a client-side web application that provides an interface to the model generation module layer 2154 or a mass storage device associated with the client layer. The user may be provided with a UI for selecting a test method and / or habitat for which biodegradability will be determined. The user may also be provided with a UI for selecting synthetic specification data. The user interface may also provide options for uploading selected data to the model generation module layer and, optionally, starting model generation.

[0085] FIG. 7 illustrates an exemplary system 700 for producing chemical products based on synthesis specifications generated in accordance with the present invention. In this example, the system includes a user interface 710 and a processor 720 associated with a control unit 740. The user interface 710 and the processor 720 may be associated with or implemented in accordance with the principles described above, and in particular may be adapted to execute a computer-implemented method for determining target functional compounds and / or synthesis specifications based on determined biodegradability, as described above. The control unit 740 is configured to receive control data generated in accordance with the present invention, for example, as described above, and in particular to receive control data generated based on synthesis specifications for functional compounds including target biodegradability. In this example, the control data is provided from a database 730, but in other examples, the control data may also be provided from a server or any other computing unit for distributing data. Vessels 750 and 752 each contain components of a chemical product, such as starting materials, catalysts, etc. Typically, there are three or more vessels, but only two are shown in this example for illustrative purposes. Valves 760 and 762 are associated with the vessels 750 and 752. Valves 750 and 752 may be controlled to dose the appropriate amount of each component into reaction vessel 770 according to the synthesis specifications. A motor 800 for agitator 780 may also be controlled by the control unit according to the synthesis specifications. An optional heater 790 may also be controlled according to the synthesis specifications. Finally, an outlet valve 810 in fluid communication with the reaction vessel may be controlled by the control unit to provide chemical products to a vessel or test system 820.

[0086] Below, we provide a more detailed example of a possible biodegradation model. In this example, a machine learning feature model is first utilized to determine molecular features that determine the biodegradability of the molecule. For example, the feature model can be a graph neural network (GNN). The GNN interprets a molecule as a graph, with the atoms of the molecule being nodes and the atomic bonds of the molecule being edges of the graph. The GNN is then configured to utilize this graph to find relevant substructures of the molecule. The feature model is then trained to convert the molecule into a vector of a predetermined number of features, e.g., 20 features, so that similar molecules are similar in their representation in these predetermined number of features. Similarity can be determined based on Tanimoto similarity, which is based on Morgan fingerprints. Further details of examples of GNNs that can be used in this case can be found in the papers "Molecular contrastive learning of representations via graph neural networks," Wang Y., Wang J., Cao Z. et al., Nat Mach Intell 4, 279-287 (2022) and "Improving Molecular Contrastive Learning via Faulty Negative Mitigation and Decomposed Fragment Contrast," Wang Y., Magar R., Liang CJ Chem.Inf.Model. 62, 11, 2713-2725 (2022). A feature vector containing the values ​​of each feature of the molecule can then be used as a digital representation of the molecule. A random forest model is then used as a biodegradation model and trained to determine biodegradation of each habitat based on the digital representation.

[0087] Both models are then trained together using public training data, e.g., from the NITE database, to perform biodegradation measurements of each molecule in each habitat. In the example provided below, over 3,000 data points are used. Measurements were performed using a standardized OECD 301 (A-F) test setup in each test habitat, with a measurement error uncertainty of approximately 10%. The molecules used for training in this exemplary training data are organic compounds with a molecular mass of less than 500 g / mol for over 90% of the molecules. As is common in machine learning, various parameters of both models are set to perform best on the available data, i.e., to predict biodegradation as accurately as possible with the data at hand. Standard machine learning training protocols are followed to ensure and test that the models can generalize to other molecules. Because both models are trained simultaneously on the same training data, a biodegradation model can also be considered to include both models and utilize digital representations, such as molecular graphs or digital representations of molecular graphs, as input.

[0088] Cross-validation can be used to evaluate the performance of the trained model, particularly the combination of the GNN model and the random forest model that map molecules to their biodegradation values. Thus, in this example, the training data used is divided into a predetermined number of data portions, for example, five portions. From the data portions, several random data portions are each used to train a model based on the respective data portion. Then, using at least one data portion, the trained model is applied to the molecules in this data portion, and the biodegradation determination results are compared with known values ​​for biodegradation. This technique makes it possible to simulate how well the model can determine the biodegradability of molecules that it has not trained on. For this exemplary model, the error averages 19%, taking into account a measurement uncertainty of 10% and a median error of 14%.

[0089] As an optional extension of the model, a method for quantifying the uncertainty of the model's decision can be implemented based on how similar each molecule is to the molecules in the training data used. Furthermore, molecules can be visualized in two-dimensional graphics based on the numerical representation from the GNN model, for example, projecting 20 feature values ​​into 2D space to enable visualization while maintaining a fairly good degree of similarity. For the model trained above, the average prediction error determined using cross-validation is 19% and the median error is 13%, specifically for the 838 molecules used as test input from the NITE database. An example is shown in the table below.

[0090] [Table 1]

[0091] In general, possible training data sources, i.e., data sources that can be used to derive training data for the respective biodegradation models, are, for example, data from the NITE database, data retrieved from literature articles, data retrieved from the Aropha dataset. Furthermore, respective measurements of biodegradation of training molecules can also be performed on a large scale on test samples using respective standard procedures and methods, for example, as described in OECD 301(A-F).

[0092] Although the biodegradation model above is described as being only a random forest model and the digital representation is determined by a GNN model, in other embodiments, the digital representation may be considered to be a molecular structure or graph or any representation showing a structure or graph, in which case the biodegradation model includes both a random forest model and a GNN, and thus refers to a two-stage machine learning model, where the output of the first model is the input of the second model.

[0093] In general, the present invention refers to a method for determining the biodegradability of, for example, a new functional compound. For example, in a first step, a digital representation of the functional compound can be provided. The digital representation can be a recipe, a structural formula, a brand name, a CAS number, etc. In an optional step, a target use of the new functional compound can be provided. In a second step, a habitat can be selected. In this context, the term "habitat" refers to the biological environment in which biodegradability is to be evaluated. Depending on the habitat, other relevant parameters can also be provided. In one embodiment, the habitat can be selected based on the target property. For example, personal care products such as shampoos are generally desired to be degraded in wastewater. As a result, in this example, automated selection selects wastewater as the habitat.

[0094] Generally, biodegradation models can be based on habitat descriptors that are important to the habitat. As a result, different biodegradation models can be selected based on the input habitat. Thus, in a preferred workflow, a biodegradation model is selected based on the habitat. After selecting a model, the model's inputs, the model may indicate that further inputs are needed, such as additional characterization parameters, and further inputs may be requested. This can be done by providing a list of the required parameters to be provided. Values ​​can be derived from digital representations of the characterization parameters, which form the inputs to the model. Based on the habitat descriptors and characterization parameters, a measure of biodegradability is determined and provided. In an optional step, a representation of biodegradability can be selected. In this case, the output is based on the selection. Potential representations of biodegradability can be one or more of: mineralization, which refers to information about whether a functional compound is completely mineralized or the time until mineralization is achieved; biotransformation, which refers to a specific property of a functional compound, such as a change in chemical structure that results in loss of toxicity, or the time until this is achieved; and half-life, which refers to the time until 50% of a functional compound is mineralized. Notable habitats are oceans, wastewater, compost, and soil. In the ocean, the following parameters may affect biodegradation: salinity, sediment, water temperature, bacterial culture, etc. In some examples, marine habitat descriptors may be stored in a database along with their geolocation. A geolocation can then be entered, and the values ​​of the parameters associated with this geolocation can be retrieved from the database. For wastewater, the following parameters may affect biodegradation: temperature, bacterial population, type of bacteria, enzyme concentration, enzymes. For soil, the following parameters may affect biodegradation: temperature, bacterial population, type of bacteria, enzyme concentration, enzymes. For compost, the following may affect biodegradation: temperature, humidity, bacterial population, type of bacteria, enzyme concentration, enzymes, compost composition.

[0095] 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.

[0096] For the processes and methods disclosed herein, the operations performed in the processes and methods may be performed in different orders. Furthermore, the outlined operations are provided only as examples, and some of the operations are optional and may be combined into fewer steps and operations, supplemented with further operations, or expanded into additional operations without detracting from the essence of the embodiments of the present disclosure.

[0097] 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.

[0098] 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.

[0099] The steps performed by one or several units or devices, such as providing the physicochemical parameters and biodegradation model of the functional compound, determining the biodegradability, providing the biodegradability, may be performed by any number of other units or devices, and these steps may be implemented as program code means of a computer program and / or as dedicated hardware.

[0100] The 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, or may be distributed in other forms, for example via the Internet or other wired or wireless telecommunications systems.

[0101] Any unit described herein may be a processing unit that is part of a classical computing system. A processing unit may include a general-purpose processor, a field programmable gate array (FPGA), an application-specific integrated circuit (ASIC), or any other dedicated circuit. Memory may be physical system memory and may be volatile, nonvolatile, or a combination of both. The term "memory" may include computer-readable storage media, such as non-volatile mass storage. If a computing system is distributed, processing power and / or storage power may also be distributed. A computing system may include multiple structures as "executable components." The term "executable component" is a structure well understood in the computing field, which may be software, hardware, or a combination thereof. For example, when implemented in software, those skilled in the art will understand that the executable component structure may include software objects, routines, methods, etc. that can be executed on a computing system. This may include both executable components in the computing system's heap or on a computer-readable storage medium. The executable component structure may reside on a computer-readable medium such that, when interpreted by one or more processors, e.g., processor threads, of the computing system, it causes the computing system to perform a function. Such structures may be directly computer readable by a processor, such as where the executable components are binary, or may be structured to be interpretable and / or compiled to generate such binary data that is directly interpretable by a processor, whether in a single step or multiple steps, for example. In other examples, the structures may be hard-coded or hard-wired logic gates implemented exclusively or nearly exclusively in hardware, such as in a field programmable gate array (FPGA), application specific integrated circuit (ASIC), or other dedicated circuitry.Thus, the term "executable component" is a term for a structure well understood by those skilled in the art of computing, whether implemented in software, hardware, or a combination. Any of the embodiments herein are described with reference to operations performed by one or more processing units of a computing system. When such operations are implemented in software, one or more processors direct the operation of the computing system in response to execution of the computer-executable instructions that make up the executable component. A computing system may also include communication channels that enable the computing system to communicate with other computing systems, for example, over a network. A "network" is defined as one or more data links that enable the transmission of electronic data between computing systems and / or modules and / or other electronic devices. When information is transferred or provided to a computing system via a network or another communications connection, e.g., either hardwired, wireless, or a combination of hardwired and wireless, the computing system properly considers the connection to be a carrier medium. A carrier medium may include a network and / or data link that may be used to carry desired program code means in the form of computer-executable instructions or data structures and that may be accessed by a general-purpose computing system or a special-purpose computing system, or a combination thereof. Although not all computing systems require a user interface, in some embodiments a computing system includes a user interface system for use in interfacing with a user. The user interface serves as an input or output mechanism to the user, for example, via a display.

[0102] Those skilled in the art will appreciate that at least portions of the present invention may be implemented in networked computing environments having many types of computing system configurations, including personal computers, desktop computers, laptop computers, message processors, handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, cellular phones, PDAs, pagers, routers, switches, data centers, wearable devices such as eyeglasses, etc. The present invention may also be implemented in distributed system environments where local and remote computing systems, linked, for example, through a network, either by hardwired data links, wireless data links, or a combination of hardwired and wireless data links, both perform tasks. In a distributed system environment, program modules may be located in both local and remote memory storage devices.

[0103] Those skilled in the art will appreciate that at least a portion of the present invention may also be implemented in a cloud computing environment. A cloud computing environment may be distributed, but this is not required. If distributed, a cloud computing environment may be distributed internationally within an organization and / or may have components held across multiple organizations. For purposes of this specification and the claims that follow, "cloud computing" is defined as a model that enables on-demand network access to a shared pool of configurable computing resources, such as networks, servers, storage, applications, and services. The definition of "cloud computing" is not limited to any of the many other advantages that may be obtained when such a model is deployed. The computing system in the figures, as described, includes various components or functional blocks that may implement various embodiments disclosed herein. The various components or functional blocks may be implemented on a local computing system or on a distributed computing system that includes elements that reside 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 fewer components than those shown in the figures, and where circumstances permit, some of the components may be combined.

[0104] Any reference signs in the claims should not be construed as limiting the scope.

[0105] The present invention relates to a method for determining biodegradability for a functional compound. A digital representation of the functional compound is provided, indicating the physicochemical properties of the functional compound. Additionally, habitats are provided, indicating habitat descriptor values ​​that affect the biodegradation of the functional compound. The habitat descriptors indicate environmental characteristics of the habitats. A biodegradation model is provided based on the habitats, and the biodegradation model is adapted to determine the biodegradability of the functional compound in each habitat, where the biodegradation model is a data-driven model parameterized with respect to the habitat so as to determine the biodegradability of the functional compound based on the physicochemical properties. The biodegradability of the functional compound is then determined based on the provided biodegradation model and the digital representation of the functional compound.

Claims

1. 1. A computer-implemented method for determining biodegradability that can be used to verify the biodegradation of a functional compound, comprising: providing a digital representation of the functional compound (210); Providing (220) biodegradation habitats, each of which exhibits a habitat descriptor value for a habitat descriptor that influences the biodegradation of a functional compound in the respective habitat, the habitat descriptor indicating an environmental characteristic of the habitat; Providing (230) a biodegradation model based on the provided biodegradation habitats, the biodegradation model being adapted to determine the biodegradability of a functional compound in each biodegradation habitat, the biodegradation model being a data-driven model parameterized with respect to the biodegradation habitat to determine the biodegradability of the functional compound based on the digital representation of the functional compound; determining (240) the biodegradability of the functional compound based on the provided biodegradation model and the digital representation of the functional compound; 20. A computer-implemented method comprising:

2. The method of claim 1 , wherein the functional compound has a molecular mass of less than 10,000 g / mol.

3. 10. The method of any one of the preceding claims, wherein the functional compound has at least one of the following properties: having an effect on a living organism, being suitable for influencing the structure, or being suitable for influencing the function of a living organism.

4. 10. The method of any one of the preceding claims, wherein the functional compound comprises at least one of the following functional groups: a carboxylic acid group, a carboxylic acid derivative group, an ether group, a hydroxyl group, a carbonyl group, an amine group, an ammonium group, an alkyl group, an alkylene group, a phenyl group, an acetal group, a ketal group, a thiol group, a sulfide group, a phosphate group, a halide group, or a combination thereof.

5. 10. The method of any one of the preceding claims, wherein the digital representation comprises or indicates a chemical structure of the functional compound, the chemical structure being related to and / or indicating one or more structural formulas and quantitative ratios of the two or more structural formulas.

6. 6. The method of claim 5, wherein the at least two structural formulas correspond to structural formulas that are related via chemical equilibrium and / or stereochemistry.

7. 10. The method of any one of the preceding claims, further comprising providing a biodegradation test method, wherein said provided biodegradation test method represents a standardized biodegradation test method for experimentally determining biodegradation of a chemical substance, and wherein said biodegradation model is further provided based on said provided biodegradation test method.

8. 1. An interface method for providing an interface, comprising: receiving as input a digital representation and a habitat via a user interface and providing said received digital representation and said habitat to a processor executing the method of any one of claims 1 to 7; Providing the determined biodegradability of the functional compound to a user via a user interface as a result, the result being received from the processor executing the method of any one of claims 1 to 7. An interface method comprising:

9. 1. A computer-implemented training method for training a data-driven based biodegradation model to parameterize said biodegradation model, comprising: Providing (310) training data associated with a given biodegradation habitat, the training data including: a) digital representations of a plurality of training functional compounds; and b) a biodegradability of the respective biodegradation habitat associated with each training functional compound; Providing a data-driven, trainable biodegradation model (320); training (330) the provided data-driven biodegradation model based on the provided training data, wherein the trained biodegradation model is adapted to determine the biodegradation of the functionalized compound based on the digital representation of the functionalized compound; providing the trained biodegradation model (340); A computer-implemented training method comprising:

10. 1. An apparatus for determining biodegradability that can be used to verify the biodegradation of a given functional compound, comprising: a digital representation providing unit (111) for providing a digital representation of said functional compound; a habitat providing unit (112) for providing biodegradation habitats, the biodegradation habitats exhibiting habitat descriptor values ​​of habitat descriptors that affect the biodegradation of functional compounds in the respective habitats, the habitat descriptors indicating environmental characteristics of the habitats; a model providing unit (113) for providing a biodegradation model based on the provided biodegradation habitats, the biodegradation model being adapted to determine the biodegradability of functional compounds in the respective biodegradation habitats, the biodegradation model being a data-driven model parameterized with respect to the biodegradation habitats for determining the biodegradability of functional compounds based on the digital representation; a determining unit (114) for determining the biodegradability of the functional compound based on the selected biodegradation model and the digital representation of the functional compound; An apparatus comprising:

11. 1. An interface device for providing an interface, comprising: an input interface unit for receiving as input a digital representation and a habitat via a user interface and for providing said received digital representation and said habitat to the apparatus of claim 10; a result interface for providing the determined biodegradability of the functional compound to a user via a user interface, the result being received from the device of claim 10; An interface device comprising:

12. 1. A training device for training a data-driven based biodegradation model to parameterize said biodegradation model, comprising: a training data providing unit (121) for providing training data associated with a predetermined biodegradation habitat, the training data including: a) digital representations of a plurality of training functional compounds; and b) biodegradability of the respective biodegradation habitat associated with each training functional compound; a trainable model providing unit (122) for providing a data-driven based trainable biodegradation model; a training unit (123) for training the provided data-driven based biodegradation model based on the provided training data, wherein the trained biodegradation model is adapted to determine the biodegradation of the functionalized compound based on a digital representation of the functionalized compound; a trained model providing unit (124) for providing said trained biodegradation model; A training device comprising:

13. A computer program product for determining the biodegradability of a given functional compound, the computer program product comprising program code means for causing an apparatus according to claim 10 to carry out the method according to any one of claims 1 to 7.

14. A computer program product for training a biodegradation model, comprising program code means for causing an apparatus according to claim 12 to carry out the method according to claim 9.

Images