Prediction of pathogenic single amino acid substitutions using molecular fragment descriptors.

MOTIVATION: Next Generation Sequencing technologies make it possible to detect rare genetic variants in individual patients. Currently, more than a dozen software and web services have been created to predict the pathogenicity of variants related with changing of amino acid residues. Despite considerable efforts in this area, at the moment there is no ideal method to classify pathogenic and harmless variants, and the assessment of the pathogenicity is often contradictory. In this article, we propose to use peptides structural formulas of proteins as an amino acid residues substitutions description, rather than a single-letter code. This allowed us to investigate the effectiveness of chemoinformatics approach to assess the pathogenicity of variants associated with amino acid substitutions. RESULTS: The structure-activity relationships analysis relying on protein-specific data and atom centric substructural multilevel neighborhoods of atoms (MNA) descriptors of molecular fragments appeared to be suitable for predicting the pathogenic effect of single amino acid variants. MNA-based Naïve Bayes classifier algorithm, ClinVar and humsavar data were used for the creation of structure-activity relationships models for 10 proteins. The performance of the models was compared with 11 different predicting tools: 8 individual (SIFT 4G, Polyphen2 HDIV, MutationAssessor, PROVEAN, FATHMM, MVP, LIST-S2, MutPred) and 3 consensus (M-CAP, MetaSVM, MetaLR). The accuracy of MNA-based method varies for the proteins (AUC: 0.631-0.993; MCC: 0.191-0.891). It was similar for both the results of comparisons with the other individual predictors and third-party protein-specific predictors. For several proteins (BRCA1, BRCA2, COL1A2, and RYR1), the performance of the MNA-based method was outstanding, capable of capturing the pathogenic effect of structural changes in amino acid substitutions. AVAILABILITY AND IMPLEMENTATION: The datasets are available as supplemental data at Bioinformatics online. A python script to convert amino acid and nucleotide sequences from single-letter codes to SD files is available at https://github.com/SmirnygaTotoshka/SequenceToSDF. The authors provide trial licenses for MultiPASS software to interested readers upon request.

TCR-Pred: A new web-application for prediction of epitope and MHC specificity for CDR3 TCR sequences using molecular fragment descriptors.

The search for the relationships between CDR3 TCR sequences and epitopes or MHC types is a challenging task in modern immunology. We propose a new approach to develop the classification models of structure-activity relationships (SAR) using molecular fragment descriptors MNA (Multilevel Neighbourhoods of Atoms) to represent CDR3 TCR sequences and the naïve Bayes classifier algorithm. We have created the freely available TCR-Pred web application (http://way2drug.com/TCR-pred/) to predict the interactions between α chain CDR3 TCR sequences and 116 epitopes or 25 MHC types, as well as the interactions between ß chain CDR3 TCR sequences and 202 epitopes or 28 MHC types. The TCR-Pred web application is based on the data (more 250 000 unique CDR3 TCR sequences) from VDJdb, McPAS-TCR, and IEDB databases and the proposed approach. The average AUC values of the prediction accuracy calculated using a 20-fold cross-validation procedure varies from 0.857 to 0.884. The created web application may be useful in studies related with T-cell profiling based on CDR3 TCR sequences.

SAV-Pred: A Freely Available Web Application for the Prediction of Pathogenic Amino Acid Substitutions for Monogenic Hereditary Diseases Studied in Newborn Screening.

Next Generation Sequencing (NGS) technologies are rapidly entering clinical practice. A promising area for their use lies in the field of newborn screening. The mass screening of newborns using NGS technology leads to the discovery of a large number of new missense variants that need to be assessed for association with the development of hereditary diseases. Currently, the primary analysis and identification of pathogenic variations is carried out using bioinformatic tools. Although extensive efforts have been made in the computational approach to variant interpretation, there is currently no generally accepted pathogenicity predictor. In this study, we used the sequence-structure-property relationships (SSPR) approach, based on the representation of protein fragments by molecular structural formula. The approach predicts the pathogenic effect of single amino acid substitutions in proteins related with twenty-five monogenic heritable diseases from the Uniform Screening Panel for Major Conditions recommended by the Advisory Committee on Hereditary Disorders in Newborns and Children. In order to create SSPR models of classification, we modified a piece of cheminformatics software, MultiPASS, that was originally developed for the prediction of activity spectra for drug-like substances. The created SSPR models were compared with traditional bioinformatic tools (SIFT 4G, Polyphen-2 HDIV, MutationAssessor, PROVEAN and FATHMM). The average AUC of our approach was 0.804 ± 0.040. Better quality scores were achieved for 15 from 25 proteins with a significantly higher accuracy for some proteins (IVD, HADHB, HBB). The best SSPR models of classification are freely available in the online resource SAV-Pred (Single Amino acid Variants Predictor).

Predicting the Impact of OTOF Gene Missense Variants on Auditory Neuropathy Spectrum Disorder.

Auditory neuropathy spectrum disorder (ANSD) associated with mutations of the OTOF gene is one of the common types of sensorineural hearing loss of a hereditary nature. Due to its high genetic heterogeneity, ANSD is considered one of the most difficult hearing disorders to diagnose. The dataset from 270 known annotated single amino acid substitutions (SAV) related to ANSD was created. It was used to estimate the accuracy of pathogenicity prediction using the known (from dbNSFP4.4) method and a new one. The new method (ConStruct) for the creation of the protein-centric classification model is based on the use of Random Forest for the analysis of missense variants in exons of the OTOF gene. A system of predictor variables was developed based on the modern understanding of the structure and function of the otoferlin protein and reflecting the location of changes in the tertiary structure of the protein due to mutations in the OTOF gene. The conservation values of nucleotide substitutions in genomes of 100 vertebrates and 30 primates were also used as variables. The average prediction of balanced accuracy and the AUC value calculated by the 5-fold cross-validation procedure were 0.866 and 0.903, respectively. The model shows good results for interpreting data from the targeted sequencing of the OTOF gene and can be implemented as an auxiliary tool for the diagnosis of ANSD in the early stages of ontogenesis. The created model, together with the results of the pathogenicity prediction of SAVs via other known accurate methods, were used for the evaluation of a manually created set of 1302 VUS related to ANSD. Based on the analysis of predicted results, 16 SAVs were selected as the new most probable pathogenic variants.

CLC-Pred 2.0: A Freely Available Web Application for In Silico Prediction of Human Cell Line Cytotoxicity and Molecular Mechanisms of Action for Druglike Compounds.

In vitro cell-line cytotoxicity is widely used in the experimental studies of potential antineoplastic agents and evaluation of safety in drug discovery. In silico estimation of cytotoxicity against hundreds of tumor cell lines and dozens of normal cell lines considerably reduces the time and costs of drug development and the assessment of new pharmaceutical agent perspectives. In 2018, we developed the first freely available web application (CLC-Pred) for the qualitative prediction of cytotoxicity against 278 tumor and 27 normal cell lines based on structural formulas of 59,882 compounds. Here, we present a new version of this web application: CLC-Pred 2.0. It also employs the PASS (Prediction of Activity Spectra for Substance) approach based on substructural atom centric MNA descriptors and a Bayesian algorithm. CLC-Pred 2.0 provides three types of qualitative prediction: (1) cytotoxicity against 391 tumor and 47 normal human cell lines based on ChEMBL and PubChem data (128,545 structures) with a mean accuracy of prediction (AUC), calculated by the leave-one-out (LOO CV) and the 20-fold cross-validation (20F CV) procedures, of 0.925 and 0.923, respectively; (2) cytotoxicity against an NCI60 tumor cell-line panel based on the Developmental Therapeutics Program's NCI60 data (22,726 structures) with different thresholds of IG50 data (100, 10 and 1 nM) and a mean accuracy of prediction from 0.870 to 0.945 (LOO CV) and from 0.869 to 0.942 (20F CV), respectively; (3) 2170 molecular mechanisms of actions based on ChEMBL and PubChem data (656,011 structures) with a mean accuracy of prediction 0.979 (LOO CV) and 0.978 (20F CV). Therefore, CLC-Pred 2.0 is a significant extension of the capabilities of the initial web application.

Towards Novel Potential Molecular Targets for Antidepressant and Antipsychotic Pharmacotherapies.

Depression and schizophrenia are two highly prevalent and severely debilitating neuropsychiatric disorders. Both conventional antidepressant and antipsychotic pharmacotherapies are often inefficient clinically, causing multiple side effects and serious patient compliance problems. Collectively, this calls for the development of novel drug targets for treating depressed and schizophrenic patients. Here, we discuss recent translational advances, research tools and approaches, aiming to facilitate innovative drug discovery in this field. Providing a comprehensive overview of current antidepressants and antipsychotic drugs, we also outline potential novel molecular targets for treating depression and schizophrenia. We also critically evaluate multiple translational challenges and summarize various open questions, in order to foster further integrative cross-discipline research into antidepressant and antipsychotic drug development.

Relationships between the Structure and Severe Drug-Induced Liver Injury for Low, Medium, and High Doses of Drugs.

Assessment of structure-activity relationships (SARs) for predicting severe drug-induced liver injury (DILI) is essential since in vivo and in vitro preclinical methods cannot detect many druglike compounds disrupting liver functions. To date, plenty of SAR models for the prediction of DILI have been developed; however, none of them considered the route of drug administration and daily dose, which may introduce significant bias into prediction results. We have created a dataset of 617 drugs with parenteral and oral administration routes and consistent information on DILI severity. We have found a clear relationship between route, dose, and DILI severity. According to SAR, nearly 40% of moderate- and non-DILI-causing drugs would cause severe DILI if they were administered at high oral doses. We have proposed the following approach to predict severe DILI. New compounds recommended to be used at low oral doses (<∼10 mg daily), or parenterally, can be considered not causing severe DILI. DILI for compounds administered at medium oral doses (∼10-100 mg daily; 22.2% of drugs under consideration) can be considered unpredictable because reasonable SAR models were not obtained due to the small size and heterogeneity of the corresponding dataset. The DILI potential of the compounds recommended to be used at high oral doses (more than ∼100 mg daily) can be estimated using SAR modeling. The balanced accuracy of the approach calculated by a 10-fold cross-validation procedure is 0.803. The developed approach can be used to estimate severe DILI for druglike compounds proposed to use at low and high oral doses or parenterally at the early stages of drug development.

Computational Prediction of Inhibitors and Inducers of the Major Isoforms of Cytochrome P450.

Human cytochrome P450 enzymes (CYPs) are heme-containing monooxygenases. This superfamily of drug-metabolizing enzymes is responsible for the metabolism of most drugs and other xenobiotics. The inhibition of CYPs may lead to drug-drug interactions and impair the biotransformation of drugs. CYP inducers may decrease the bioavailability and increase the clearance of drugs. Based on the freely available databases ChEMBL and PubChem, we have collected over 70,000 records containing the structures of inhibitors and inducers together with the IC50 values for the inhibitors of the five major human CYPs: 1A2, 3A4, 2D6, 2C9, and 2C19. Based on the collected data, we developed (Q)SAR models for predicting inhibitors and inducers of these CYPs using GUSAR and PASS software. The developed (Q)SAR models could be applied for assessment of the interaction of novel drug-like substances with the major human CYPs. The created (Q)SAR models demonstrated reasonable accuracy of prediction. They have been implemented in the web application P450-Analyzer that is freely available via the Internet.

Pyridine Carboxamides Based on Sulfobetaines: Design, Reactivity, and Biological Activity.

The synthesis of the products of the 1,3-propanesultone ring opening during its interaction with amides of pyridinecarboxylic acids has been carried out. The dependence of the yield of the reaction products on the position (ortho-, meta-, para-) of the substituent in the heteroaromatic fragment and temperature condition was revealed. In contrast to the meta- and para-substituted substrates, the reaction involving ortho-derivatives at the boiling point of methanol unexpectedly led to the formation of a salt. On the basis of spectroscopic, X-Ray, and quantum-chemical calculation data, a model of the transition-state, as well as a mechanism for this alkylation reaction of pyridine carboxamides with sultone were proposed in order to explain the higher yields obtained with the nicotinamide and its N-methyl analog compared to ortho or meta parents. Based on the analysis of ESP maps, the positions of the binding sites of reagents with a potential complexing agent in space were determined. The in silico evaluation of possible biological activity showed that the synthetized compounds revealed some promising pharmacological effects and low acute toxicity.

A Novel Phenylpyrrolidine Derivative: Synthesis and Effect on Cognitive Functions in Rats with Experimental Ishemic Stroke.

We performed an in silico, in vitro, and in vivo assessment of a potassium 2-[2-(2-oxo-4-phenylpyrrolidin-1-yl) acetamido]ethanesulfonate (compound 1) as a potential prodrug for cognitive function improvement in ischemic brain injury. Using in silico methods, we predicted the pharmacological efficacy and possible safety in rat models. In addition, in silico data showed neuroprotective features of compound 1, which were further supported by in vitro experiments in a glutamate excitotoxicity-induced model in newborn rat cortical neuron cultures. Next, we checked whether compound 1 is capable of crossing the blood-brain barrier in intact and ischemic animals. Compound 1 improved animal behavior both in intact and ischemic rats and, even though the concentration in intact brains was low, we still observed a significant anxiety reduction and activity escalation. We used molecular docking and molecular dynamics to support our hypothesis that compound 1 could affect the AMPA receptor function. In a rat model of acute focal cerebral ischemia, we studied the effects of compound 1 on the behavior and neurological deficit. An in vivo experiment demonstrated that compound 1 significantly reduced the neurological deficit and improved neurological symptom regression, exploratory behavior, and anxiety. Thus, here, for the first time, we show that compound 1 can be considered as an agent for restoring cognitive functions.

Assessment of the cardiovascular adverse effects of drug-drug interactions through a combined analysis of spontaneous reports and predicted drug-target interactions.

Adverse drug effects (ADEs) are one of the leading causes of death in developed countries and are the main reason for drug recalls from the market, whereas the ADEs that are associated with action on the cardiovascular system are the most dangerous and widespread. The treatment of human diseases often requires the intake of several drugs, which can lead to undesirable drug-drug interactions (DDIs), thus causing an increase in the frequency and severity of ADEs. An evaluation of DDI-induced ADEs is a nontrivial task and requires numerous experimental and clinical studies. Therefore, we developed a computational approach to assess the cardiovascular ADEs of DDIs. This approach is based on the combined analysis of spontaneous reports (SRs) and predicted drug-target interactions to estimate the five cardiovascular ADEs that are induced by DDIs, namely, myocardial infarction, ischemic stroke, ventricular tachycardia, cardiac failure, and arterial hypertension. We applied a method based on least absolute shrinkage and selection operator (LASSO) logistic regression to SRs for the identification of interacting pairs of drugs causing corresponding ADEs, as well as noninteracting pairs of drugs. As a result, five datasets containing, on average, 3100 potentially ADE-causing and non-ADE-causing drug pairs were created. The obtained data, along with information on the interaction of drugs with 1553 human targets predicted by PASS Targets software, were used to create five classification models using the Random Forest method. The average area under the ROC curve of the obtained models, sensitivity, specificity and balanced accuracy were 0.837, 0.764, 0.754 and 0.759, respectively. The predicted drug targets were also used to hypothesize the potential mechanisms of DDI-induced ventricular tachycardia for the top-scoring drug pairs. The created five classification models can be used for the identification of drug combinations that are potentially the most or least dangerous for the cardiovascular system.

Prediction of Protein-ligand Interaction Based on Sequence Similarity and Ligand Structural Features.

Computationally predicting the interaction of proteins and ligands presents three main directions: the search of new target proteins for ligands, the search of new ligands for targets, and predicting the interaction of new proteins and new ligands. We proposed an approach providing the fuzzy classification of protein sequences based on the ligand structural features to analyze the latter most complicated case. We tested our approach on five protein groups, which represented promised targets for drug-like ligands and differed in functional peculiarities. The training sets were built with the original procedure overcoming the data ambiguity. Our study showed the effective prediction of new targets for ligands with an average accuracy of 0.96. The prediction of new ligands for targets displayed the average accuracy 0.95; accuracy estimates were close to our previous results, comparable in accuracy to those of other methods or exceeded them. Using the fuzzy coefficients reflecting the target-to-ligand specificity, we provided predicting interactions for new proteins and new ligands; the obtained accuracy values from 0.89 to 0.99 were acceptable for such a sophisticated task. The protein kinase family case demonstrated the ability to account for subtle features of proteins and ligands required for the specificity of protein-ligand interaction.

A Computational Approach for the Prediction of Treatment History and the Effectiveness or Failure of Antiretroviral Therapy.

Human Immunodeficiency Virus Type 1 (HIV-1) infection is associated with high mortality if no therapy is provided. Currently, the treatment of an HIV-1 positive patient requires that several drugs should be taken simultaneously. The resistance of the virus to an antiretroviral drug may lead to treatment failure. Our approach focuses on predicting the exposure of a particular viral variant to an antiretroviral drug or drug combination. It also aims at the prediction of drug treatment success or failure. We utilized nucleotide sequences of HIV-1 encoding protease and reverse transcriptase to perform such types of prediction. The PASS (Prediction of Activity Spectra for Substances) algorithm based on the naive Bayesian classifier was used to make a prediction. We calculated the probability of whether a sequence belonged (P1) or did not belong (P0) to the class associated with exposure of the viral sequence to the set of drugs that can be associated with resistance to the set of drugs. The accuracy calculated as the average Area Under the ROC (Receiver Operating Characteristic) Curve (AUC/ROC) for classifying exposure of the sequence to the HIV-1 protease inhibitors was 0.81 (±0.07), and for HIV-1 reverse transcriptase, it was 0.83 (±0.07). To predict cases of treatment effectiveness or failure, we used P1 and P0 values, obtained in PASS, along with the binary vector constructed based on short nucleotide descriptors and the applied random forest classifier. Average AUC/ROC prediction accuracy for the prediction of treatment effectiveness or failure for the combinations of HIV-1 protease inhibitors was 0.82 (±0.06) and of HIV-1 reverse transcriptase was 0.76 (±0.09).

ROSC-Pred: web-service for rodent organ-specific carcinogenicity prediction.

Motivation: Identification of rodent carcinogens is an important task in risk assessment of chemicals. SAR methods were proposed to reduce the number of animal experiments. Most of these methods ignore information about organ-specificity of tumorigenesis. Our study was aimed at the creation of classification models and a freely available online service for prediction of rodent carcinogens considering the species (rats, mice), sex and tissue-specificity from structural formula of compounds. Results: The data from Carcinogenic Potency Database for 1011 organic compounds evaluated on the standard two-year rodent carcinogenicity bioassay was used for the creation of training sets. Structure-activity relationships models for prediction of rodent organ-specific carcinogenicity were created by PASS software, which was based on Bayesian-like approach and Multilevel Neighborhoods of Atoms descriptors. The average prediction accuracy for training sets calculated by leave-one-out and 10-fold cross-validation was 79 and 78.2%, respectively. Availability and implementation: Freely available on the web at http://www.way2drug.com/ROSC. Contact: alexey.lagunin@ibmc.msk.ru. Supplementary information: Supplementary data are available at Bioinformatics online.

Rational Use of Heterogeneous Data in Quantitative Structure-Activity Relationship (QSAR) Modeling of Cyclooxygenase/Lipoxygenase Inhibitors.

Numerous studies have been published in recent years with acceptable quantitative structure-activity relationship (QSAR) modeling based on heterogeneous data. In many cases, the training sets for QSAR modeling were constructed from compounds tested by different biological assays, contradicting the opinion that QSAR modeling should be based on the data measured by a single protocol. We attempted to develop approaches that help to determine how heterogeneous data should be used for the creation of QSAR models on the basis of different sets of compounds tested by different experimental methods for the same target and the same endpoint. To this end, more than 100 QSAR models for the IC50 values of ligands interacting with cyclooxygenase 1,2 (COX) and seed lipoxygenase (LOX), obtained from ChEMBL database were created using the GUSAR software. The QSAR models were tested on the external set, including 26 new thiazolidinone derivatives, which were experimentally tested for COX-1,2/LOX inhibition. The IC50 values of the derivatives varied from 89 µM to 26 µM for LOX, from 200 µM to 0.018 µM for COX-1, and from 210 µM to 1 µM for COX-2. This study showed that the accuracy of the models is dependent on the distribution of IC50 values of low activity compounds in the training sets. In the most cases, QSAR models created based on the combined training sets had advantages in comparison with QSAR models, based on a single publication. We introduced a new method of combination of quantitative data from different experimental studies based on the data of reference compounds, which was called "scaling".

AntiBac-Pred: A Web Application for Predicting Antibacterial Activity of Chemical Compounds.

Discovery of new antibacterial agents is a never-ending task of medicinal chemistry. Every new drug brings significant improvement to patients with bacterial infections, but prolonged usage of antibacterials leads to the emergence of resistant strains. Therefore, novel active structures with new modes of action are required. We describe a web application called AntiBac-Pred aimed to help users in the rational selection of the chemical compounds for experimental studies of antibacterial activity. This application is developed using antibacterial activity data available in ChEMBL and PASS software. It allows users to classify chemical structures of interest into growth inhibitors or noninhibitors of 353 different bacteria strains, including both resistant and nonresistant ones.

Prediction of Protein-Ligand Interaction Based on the Positional Similarity Scores Derived from Amino Acid Sequences.

The affinity of different drug-like ligands to multiple protein targets reflects general chemical-biological interactions. Computational methods estimating such interactions analyze the available information about the structure of the targets, ligands, or both. Prediction of protein-ligand interactions based on pairwise sequence alignment provides reasonable accuracy if the ligands' specificity well coincides with the phylogenic taxonomy of the proteins. Methods using multiple alignment require an accurate match of functionally significant residues. Such conditions may not be met in the case of diverged protein families. To overcome these limitations, we propose an approach based on the analysis of local sequence similarity within the set of analyzed proteins. The positional scores, calculated by sequence fragment comparisons, are used as input data for the Bayesian classifier. Our approach provides a prediction accuracy comparable or exceeding those of other methods. It was demonstrated on the popular Gold Standard test sets, presenting different sequence heterogeneity and varying from the group, including different protein families to the more specific groups. A reasonable prediction accuracy was also found for protein kinases, displaying weak relationships between sequence phylogeny and inhibitor specificity. Thus, our method can be applied to the broad area of protein-ligand interactions.

Prediction of Severity of Drug-Drug Interactions Caused by Enzyme Inhibition and Activation.

Drug-drug interactions (DDIs) severity assessment is a crucial problem because polypharmacy is increasingly common in modern medical practice. Many DDIs are caused by alterations of the plasma concentrations of one drug due to another drug inhibiting and/or inducing the metabolism or transporter-mediated disposition of the victim drug. Accurate assessment of clinically relevant DDIs for novel drug candidates represents one of the significant tasks of contemporary drug research and development and is important for practicing physicians. This work is a development of our previous investigations and aimed to create a model for the severity of DDIs prediction. PASS program and PoSMNA descriptors were implemented for prediction of all five classes of DDIs severity according to OpeRational ClassificAtion (ORCA) system: contraindicated (class 1), provisionally contraindicated (class 2), conditional (class 3), minimal risk (class 4), no interaction (class 5). Prediction can be carried out both for known drugs and for new, not yet synthesized substances using only their structural formulas. Created model provides an assessment of DDIs severity by prediction of different ORCA classes from the first most dangerous class to the fifth class when DDIs do not take place in the human organism. The average accuracy of DDIs class prediction is about 0.75.

ADVERPred-Web Service for Prediction of Adverse Effects of Drugs.

Application of structure-activity relationships (SARs) for the prediction of adverse effects of drugs (ADEs) has been reported in many published studies. Training sets for the creation of SAR models are usually based on drug label information which allows for the generation of data sets for many hundreds of drugs. Since many ADEs may not be related to drug consumption, one of the main problems in such studies is the quality of data on drug-ADE pairs obtained from labels. The information on ADEs may be included in three sections of the drug labels: "Boxed warning," "Warnings and Precautions," and "Adverse reactions." The first two sections, especially Boxed warning, usually contain the most frequent and severe ADEs that have either known or probable relationships to drug consumption. Using this information, we have created manually curated data sets for the five most frequent and severe ADEs: myocardial infarction, arrhythmia, cardiac failure, severe hepatotoxicity, and nephrotoxicity, with more than 850 drugs on average for each effect. The corresponding SARs were built with PASS (Prediction of Activity Spectra for Substances) software and had balanced accuracy values of 0.74, 0.7, 0.77, 0.67, and 0.75, respectively. They were implemented in a freely available ADVERPred web service ( http://www.way2drug.com/adverpred/ ), which enables a user to predict five ADEs based on the structural formula of compound. This web service can be applied for estimation of the corresponding ADEs for hits and lead compounds at the early stages of drug discovery.

MetaTox: Web Application for Predicting Structure and Toxicity of Xenobiotics' Metabolites.

A new freely available web-application MetaTox ( http://www.way2drug.com/mg ) for prediction of xenobiotic's metabolism and calculation toxicity of metabolites based on the structural formula of chemicals has been developed. MetaTox predicts metabolites, which are formed by nine classes of reactions (aliphatic and aromatic hydroxylation, N- and O-glucuronidation, N-, S- and C-oxidation, and N- and O-dealkylation). The calculation of probability for generated metabolites is based on analyses of "structure-biotransformation reactions" and "structure-modified atoms" relationships using a Bayesian approach. Prediction of LD50 values is performed by GUSAR software for the parent compound and each of the generated metabolites using quantitative structure-activity relationahip (QSAR) models created for acute rat toxicity with the intravenous type of administration.