Leveraging complementary computational models for prioritizing chemicals of developmental and reproductive toxicity concern: an example of food contact materials

Incorporating Tissue-Specific Gene Expression Data to Improve Chemical-Disease Inference of in Silico Toxicogenomics Methods

In silico toxicogenomics methods are resource- and time-efficient approaches for inferring chemical-protein-disease associations with potential mechanism information for exploring toxicological effects. However, current in silico toxicogenomics systems make inferences based on only chemical-protein interactions without considering tissue-specific gene/protein expressions. As a result, inferred diseases could be overpredicted with false positives. In this work, six tissue-specific expression datasets of genes and proteins were collected from the Expression Atlas. Genes were then categorized into high, medium, and low expression levels in a tissue- and dataset-specific manner. Subsequently, the tissue-specific expression datasets were incorporated into the chemical-protein-disease inference process of our ChemDIS system by filtering out relatively low-expressed genes. By incorporating tissue-specific gene/protein expression data, the enrichment rate for chemical-disease inference was largely improved with up to 62.26% improvement. A case study of melamine showed the ability of the proposed method to identify more specific disease terms that are consistent with the literature. A user-friendly user interface was implemented in the ChemDIS system. The methodology is expected to be useful for chemical-disease inference and can be implemented for other in silico toxicogenomics tools.

Thyroid under siege: Unravelling the toxic impact of real-life metal mixture exposures in Wistar rats

This study explored the effect of a toxic metal(oid) mixture (cadmium, lead, arsenic, mercury, chromium, and nickel) on thyroid function in Wistar rats exposed for 28 or 90 days. Dose levels were determined based on prior human-biomonitoring investigation. The experiment included control (male/female rats, 28 and 90 days) and treated groups, reflecting the lower confidence limit of the Benchmark Dose (BMDL) for hormone levels (M1/F1, 28 and 90 days), median concentrations (M2/F2, 28 and 90 days), 95th percentile concentrations (M3/F3, 28 and 90 days) measured in a human study, and reference values for individual metals extracted from the literature (M4/F4, 28 days only). Blood and thyroid gland samples were collected at the experimental termination. Serum TSH, fT3, fT4, T3, and T4 levels were measured, and SPINA-GT and SPINA-GD parameters were calculated. In silico analysis, employing the Comparative Toxicogenomic Database and ToppGene Suite portal, aimed to reveal molecular mechanisms underlying the observed effects. Results showed greater sensitivity in the female rats, with significant effects observed at lower doses. Subacute exposure increased TSH, fT3, and T3 levels in females, while subchronic exposure in males decreased TSH and fT3 levels and increased fT4. Subacute exposure induced changes even at allegedly safe doses, emphasizing potential health risks. Histological abnormalities were observed in all the treated groups. In silico findings suggested that toxic metal exposure contributes to thyroid disorders via oxidative stress, disruption of micronutrients, interference with hormone synthesis, and gene expression dysregulation. These results indicate that seemingly safe doses in single-substance research can adversely affect thyroid structure and function when administered as a mixture. These findings highlight the complex impact of toxic metal exposure on thyroid health, emphasizing that adhering to accepted safety limits for single-substance research fails to account for adverse effects on thyroid structure and function upon exposures to metal mixtures.

Endocrine disruptors in e-waste dismantling dust: In silico prediction of mixture-induced reproductive toxicity mechanisms

The constant exposure of humans to a mixture of low doses of toxic substances, emerging from the daily emission of toxic dust containing various metals and organic compounds in electrical and electronic waste (e-waste) recycling areas, poses potential harmful effects on health and the environment. While individually recognized as endocrine disruptors affecting hormonal balance, the combined impact of these toxic substances in a mixture remains insufficiently explored, particularly in relation to reproductive health. Thus, the aim of this in silico analysis was to: (i) assess the relationship between the exposure to a mixture of DBDE, DBDPE, TBBPA, Pb, Cd and Ni and development of male and female reproductive system disorders; and (ii) demonstrate the ability of in silico toxicogenomic tools in revealing the potential molecular mechanisms involved in the mixture toxicity. As the main data-mining tool, Comparative Toxicogenomics Database (CTD) was used, along with the ToppGene Suite portal and GeneMANIA online server. Our analysis identified 5 genes common to all the investigated substances and linked to reproductive system disorders. Notably, the most prominent interactions among these genes were physical interactions (77.64 %). Pathway enrichment analysis identified oxidative stress response as the central disrupted molecular pathway linked to reproductive pathology in the investigated mixture, while our chemical-phenotype CTD analysis uncovered additional affected pathways - apoptosis, hormonal regulation, and developmental functions. These findings highlight an increased risk of reproductive system disorders associated with the exposure to the investigated mixture of toxic substances in electronic waste recycling areas, emphasizing the urgent need for attention to address this environmental health concern. Hence, future laboratory studies should prioritize investigating the specific genes and common mechanisms identified in this study.

Abnormality of mussel in the early developmental stages induced by graphene and triphenyl phosphate: In silico toxicogenomic data-mining, in vivo, and toxicity pathway-oriented approach

Increasing number of complex mixtures of organic pollutants in coastal area (especially for nanomaterials and micro/nanoplastics associated chemicals) threaten aquatic ecosystems and their joint hazards are complex and demanding tasks. Mussels are the most sensitive marine faunal groups in the world, and their early developmental stages (embryo and larvae) are particularly susceptible to environmental contaminants, which can distinguish the probable mechanisms of mixture-induced growth toxicity. In this study, the potential critical target and biological processes affected by graphene and triphenyl phosphate (TPP) were developed by mining public toxicogenomic data. And their combined toxic effects were verified by toxicological assay at early developmental stages in filter-feeding mussels (embryo and larvae). It showed that interactions among graphene/TPP with 111 genes (ABCB1, TP53, SOD, CAT, HSP, etc.) affected phenotypes along conceptual framework linking these chemicals to developmental abnormality endpoints. The PPAR signaling pathway, monocarboxylic acid metabolic process, regulation of lipid metabolic process, response to oxidative stress, and gonad development were noted as the key molecular pathways that contributed to the developmental abnormality. Enriched phenotype analysis revealed biological processes (cell proliferation, cell apoptosis, inflammatory response, response to oxidative stress, and lipid metabolism) affected by the investigated mixture. Combined, our results supported that adverse effects induced by contaminants/ mixture could not only be mediated by single receptor signaling or be predicted by the simple additive effect of contaminants. The results offer a framework for better comprehending the developmental toxicity of environmental contaminants in mussels and other invertebrate species, which have considerable potential for hazard assessment of coastal mixture.

An in silico insight on the mechanistic aspects of gelsenicine toxicity: A reverse screening study pointing to the possible involvement of acetylcholine binding receptor

Gelsedine-type alkaloids are highly toxic plant secondary metabolites produced by shrubs belonging to the Gelsemium genus. Gelsenicine is one of the most concerning gelsedine-type alkaloids with a lethal dose lower than 1 mg/Kg in mice. Several reported episodes of poisoning in livestock and fatality cases in humans due to the usage of Gelsemium plants extracts were reported. Also, gelsedine-type alkaloids were found in honey constituting a potential food safety issue. However, their toxicological understanding is scarce and the molecular mechanism underpinning their toxicity needs further investigations. In this context, an in silico approach based on reverse screening, docking and molecular dynamics successfully identified a possible gelsenicine biological target shedding light on its toxicodynamics. In line with the available crystallographic data, it emerged gelsenicine could target the acetylcholine binding protein possibly acting as a partial agonist against α7 nicotinic acetylcholine receptor (AChR). Overall, these results agreed with evidence previously reported and prioritized AChR for further dedicated analysis.

Machine learning for predicting chemical migration from food packaging materials to foods

Food contact chemicals (FCCs) can migrate from packaging materials to food posing an issue of exposure to FCCs of toxicity concern. Compared to costly experiments, computational methods can be utilized to assess the migration potentials for various migration scenarios for further experimental investigation that can potentially accelerate the migration assessment. This study developed a nonlinear machine learning method utilizing chemical properties, material type, food type and temperature to predict chemical migration from package to food. Nine nonlinear algorithms were evaluated for their prediction performance. The ensemble model leveraging multiple algorithms provides state-of-the-art performance that is much better than previous linear regression models. The developed prediction models were subsequently applied to profile the migration potential of FCCs of high toxicity concern. The models are expected to be useful for accelerating the assessment of migration of FCCs from package to foods.

Testing sulforaphane as a strategy against toxic chemicals of public health concern by toxicogenomic data analysis: Friend or foe at the gene level - Colorectal carcinoma case study

Toxic metals (cadmium (Cd), lead (Pb), mercury (Hg) and arsenic (As)) and plastificators (bis (2 - ethylhexyl) phthalate (DEHP), dibutyl phthalate (DBP)) and bisphenol A (BPA)) have been suggested to aid in colorectal carcinoma (CRC) advancement. Sulforaphane (SFN), isothiocyanate from cruciferous vegetables, diminishes chemical carcinogenesis susceptibility, but has been shown to act as a friend or a foe depending on various factors. By conducting the mechanistic toxicogenomic data mining approach, this research aimed to determine if SFN can alleviate toxic-metal and/or phthalate/BPA mixture-induced CRC at the gene level. Comparative Toxicogenomics Database, ToppGene Suite portal, Cytoscape software, InteractiVenn and Gene Expression Omnibus (GEO) database (GEO2R tool) was used. Among the mutual genes for all the investigated substances, SFN had a protective impact only through PTGS2. Other proposed protective SFN-targets included ABCA1, ALDH2, BMP2, DPYD, MYC, SLCO2A1, and SOD2, only in the case of phthalates/BPA exposure. The only additional gene relevant for SFN protection against the toxic metal mixture-induced CRC was ABCB1. Additionally, the majority of the top 15 molecular pathways extracted for SFN impact on phthalate and BPA mixture-linked CRC development were directly linked with cancer development, which was not the case with the toxic metal mixture. The current research has indicated that SFN is a more effective chemoprotective agent against CRC induced by phthalates/BPA mixture than by toxic-metal mixture. It has also presented the value of computational methods as a simple tool for directing further research, selecting appropriate biomarkers and exploring the mechanisms of toxicity.

Toxicogenomic analysis of physiologically important metals: An integrated in silico approach

Biologically important metals regulate cellular homeostasis in living systems. Anthropogenic exposure to these metals can cause adverse effects, including an increased incidence of diseases in humans such as cancer, lung, and cardiovascular defects. However, the effects of metals and the common genes/signaling pathways involved in metal toxicity have not been elucidated. Hence, the present study used toxicogenomic data mining with the comparative toxicogenomics database to explore the impact of these metals. The metals were categorized into transition, alkali, and alkali earth. The common genes were identified and subjected to functional enrichment analysis. Further, gene-gene and protein-protein interactions were assessed. Also, the top ten transcription factors and miRNAs that regulate the genes were identified. The phenotypes and diseases that have increased incidence upon alterations of these genes were detected. Overall, we were able to identify IL1B and SOD2 as the common genes, along with the AGE-RAGE signaling pathway in diabetic complications as the common pathway altered. Enriched genes and pathways specific to each metal category were also found. Further, we identified heart failure as the major diseases that have increased the incidence of these metals' exposure. In conclusion, exposure to essential metals might cause adverse effects via inflammation and oxidative stress.

State of the Art of Genomic Technology in Toxicology: A Review

The rapid growth of genomics techniques has revolutionized and impacted, greatly and positively, the knowledge of toxicology, ushering it into a "new era": the era of genomic technology (GT). This great advance permits us to analyze the whole genome, to know the gene response to toxicants and environmental stressors, and to determine the specific profiles of gene expression, among many other approaches. The aim of this work was to compile and narrate the recent research on GT during the last 2 years (2020-2022). A literature search was managed using the PubMed and Medscape interfaces on the Medline database. Relevant articles published in peer-reviewed journals were retrieved and their main results and conclusions are mentioned briefly. It is quite important to form a multidisciplinary taskforce on GT with the aim of designing and implementing a comprehensive, collaborative, and a strategic work plan, prioritizing and assessing the most relevant diseases, so as to decrease human morbimortality due to exposure to environmental chemicals and stressors.

In silico prediction of parkinsonian motor deficits-related neurotoxicants based on the adverse outcome pathway concept

Exposure to neurotoxicants has been associated with Parkinson's disease (PD). Limited by the clinical variation in the signs and symptoms as well as the slow disease progression, the identification of parkinsonian neurotoxicants relies on animal models. Here, we propose an innovative in silico model for the prediction of parkinsonian neurotoxicants. The model was designed based on a validated adverse outcome pathway (AOP) for parkinsonian motor deficits initiated from the inhibition of mitochondrial complex I. The model consists of a molecular docking model for mitochondrial complex I protein to predict the molecular initiating event and a neuronal cytotoxicity Quantitative Structure-Activity Relationships (QSAR) model to predict the cellular outcome of the AOP. Four known PD-related complex I inhibitors and four non-neurotoxic chemicals were utilized to develop the threshold of the models and to validate the model, respectively. The integrated model showed 100% specificity in ruling out the non-neurotoxic chemicals. The screening of 41 neurotoxicants and complex I inhibitors with the model resulted in 16 chemicals predicted to induce parkinsonian disorder through the molecular initiating event of mitochondrial complex I inhibition. Five of them, namely cyhalothrin, deguelin, deltamethrin, diazepam, and permethrin, are cases with direct evidence linking them to parkinsonian motor deficit-related signs and symptoms. The neurotoxicant prediction model for parkinsonian motor deficits based on the AOP concept may be useful in prioritizing chemicals for further evaluations on PD potential.

Ensemble learning for predicting ex vivo human placental barrier permeability

Background

The placental barrier protects the fetus from exposure to some toxicants and is vital for drug development and risk assessment of environmental chemicals. However, in vivo experiments for assessing the placental barrier permeability of chemicals is not ethically acceptable. Although ex vivo placental perfusion methods provide good alternatives for the assessment of placental barrier permeability, the application to a large number of test chemicals could be time- and resource-consuming. Computational prediction models for ex vivo placental barrier permeability are therefore desirable.

Methods

A total of 87 chemicals and corresponding 1444 physicochemical properties were divided into training and test datasets. Three types of algorithms including linear regression, random forest, and ensemble models were applied to develop prediction models for ex vivo placental barrier permeability.

Results

Among the tested models, the ensemble model integrating the previous two methods performed best for predicting ex vivo human placental barrier permeability with correlation coefficients of 0.887 and 0.825 when considering the applicability domain. An additional test on seven newly curated chemicals from the literature showed a good correlation coefficient of 0.879 which was further improved to 0.921 by considering the variation of experiments.

Conclusion

In this study, the first valid predicting model for ex vivo human placental barrier permeability was developed following the OECD guideline. The model is expected to be useful for assessing the human placental barrier permeability and can be integrated with developmental toxicity prediction models for investigating the toxic effects of chemicals on the fetus.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-022-04937-y.

Joint impact of key air pollutants on COVID-19 severity: prediction based on toxicogenomic data analysis

Considering that some researchers point to a possible influence of air pollution on COVID-19 transmission, severity, and death rate, the aim of our in silico study was to determine the relationship between the key air pollutants [sulphur dioxide (SO), carbon monoxide (CO), ₂particulate matter (PM_x), nitrogen dioxide (NO₂), and ozone (O₃)] and COVID-19 complications using the publicly available toxicogenomic analytical and prediction tools: (i) Comparative Toxicogenomic Database (CTD) to identify genes common to air pollutants and COVID-19 complications; (ii) GeneMANIA to construct a network of these common and related genes; (iii) ToppGene Suite to extract the most important biological processes and molecular pathways; and (iv) DisGeNET to search for the top gene-disease pairs. SO₂, CO, PM_x, NO₂, and O₃ interacted with 6, 6, 18, 9, and 12 COVID-19-related genes, respectively. Four of these are common for all pollutants (IL10, IL6, IL1B, and TNF) and participate in most (77.64 %) physical interactions. Further analysis pointed to cytokine binding and cytokine-mediated signalling pathway as the most important molecular function and biological process, respectively. Other molecular functions and biological processes are mostly related to cytokine activity and inflammation, which might be connected to the cytokine storm and resulting COVID-19 complications. The final step singled out the link between the CEBPA gene and acute myelocytic leukaemia and between TNFRSF1A and TNF receptor-associated periodic fever syndrome. This indicates possible complications in COVID-19 patients suffering from these diseases, especially those living in urban areas with poor air quality.

Ecotoxicological risk ranking of 19 metals in the lower Yangtze River of China based on their threats to aquatic wildlife

With thousands of chemicals discharged into the aquatic environment, it is necessary to identify those that are likely to be having the greatest impact on wildlife to better protect the ecosystem. A risk ranking approach was developed to compare the ecotoxicological risk of chemicals on aquatic wildlife with a wide range of environmental measurement data and ecotoxicity data. Nineteen metals including some rarely monitored ones including antimony (Sb), molybdenum (Mo), cobalt (Co), vanadium (V), titanium (Ti) and thallium (Tl) in the lower Yangtze River were risk ranked as a case study. The risk ranking approach was conducted in three tiers: general risk ranking, lethal effects vs. non-lethal effects risk ranking, and species group-specific risk ranking. Iron, copper and titanium were identified as being of greatest concern. The contamination of iron, zinc, copper and nickel was widespread and may have already harmed wildlife according to the overlap between ecotoxicity and monitored levels. Based on this analysis, the risk from copper and some rarely monitored metals (titanium and boron) may have been underestimated. Greater efforts to reduce copper, iron and titanium contamination could make an important difference to the health of Chinese freshwater organisms in the Yangtze River.

A machine learning-driven approach for prioritizing food contact chemicals of carcinogenic concern based on complementary in silico methods

Carcinogenicity is one of the most critical endpoints for the risk assessment of food contact chemicals (FCCs). However, the carcinogenicity of FCCs remains insufficiently investigated. To fill the data gap, the application of standard experimental methods for identifying chemicals of carcinogenic concerns from a large set of FCCs is impractical due to their resource-intensive nature. In contrast, computational methods provide an efficient way to quickly screen chemicals with carcinogenic potential for subsequent experimental validation. Since every model was developed based on a limited number of training samples, the use of single models for carcinogenicity assessment may not cover the complex mechanisms of carcinogenesis. This study proposed a novel machine learning-based weight-of-evidence (WoE) model for prioritizing chemical carcinogenesis. The WoE model can nonlinearly integrate complementary computational methods of structural alerts, quantitative structure-activity relationship models and in silico toxicogenomics models into a WoE-score. Compared to the best single method, the WoE model gained 8% and 19.7% improvement in the area under the receiver operating characteristic curve (AUC) value and chemical coverage, respectively. The prioritization of 1623 FCCs concludes 44 chemicals of high carcinogenic concern. The machine learning-based WoE approach provides a fast and comprehensive way for prioritizing chemicals of carcinogenic concern.

Curation of cancer hallmark-based genes and pathways for in silico characterization of chemical carcinogenesis

Exposure to toxic substances in the environment is one of the most important causes of cancer. However, the time-consuming process for the identification and characterization of carcinogens is not applicable to a huge amount of testing chemicals. The data gaps make the carcinogenic risk uncontrollable. An efficient and effective way of prioritizing chemicals of carcinogenic concern with interpretable mechanism information is highly desirable. This study presents a curation work for genes and pathways associated with 11 hallmarks of cancer (HOCs) reported by the Halifax Project. To demonstrate the usefulness of the curated HOC data, the interacting HOC genes and affected HOC pathways of chemicals of the three carcinogen lists from IARC, NTP and EPA were analyzed using the in silico toxicogenomics ChemDIS system. Results showed that a higher number of affected HOCs were observed for known carcinogens than the other chemicals. The curated HOC data is expected to be useful for prioritizing chemicals of carcinogenic concern.

Database URL: The HOC database is available at https://github.com/hocdb-KMU-TMU/hocdb and the website of Database journal as Supplementary Data.

Elucidating the influence of environmentally relevant toxic metal mixture on molecular mechanisms involved in the development of neurodegenerative diseases: In silico toxicogenomic data-mining

This in silico toxicogenomic analysis aims to: (i) testify the hypothesis about the influence of the environmentally relevant toxic metals (lead, methylmercury (organic form of mercury), cadmium and arsenic) on molecular mechanisms involved in amyotrophic lateral sclerosis (ALS), Parkinson's Disease (PD) and Alzheimer's disease (AD) development; and (ii) demonstrate the capability of in silico toxicogenomic data-mining for distinguishing the probable mechanisms of mixture-induced toxic effects. The Comparative Toxicogenomics Database (CTD; http://ctd. mdibl.org) and Cytoscape software were used as the main data-mining tools in this analysis. The results have shown that there were 7, 13 and 14 common genes for all the metals present in the mixture for each of the selected neurodegenerative disease (ND), respectively: ALS, PD and AD. Physical interactions (68.18%) were the most prominent interactions between the genes extracted for ALS, co-expression (60.85%) for PD and interactions predicted by the server (44.30%) for AD. SOD2 gene was noted as the mutual gene for all the selected ND. Oxidative stress, folate metabolism, vitamin B12, AGE-RAGE, apoptosis were noted as the key disrupted molecular pathways that contribute to the neurodegenerative disease's development. Gene ontology analysis revealed biological processes affected by the investigated mixture (glutathione metabolic process was listed as the most important for ALS, cellular response to toxic substance for PD, and neuron death for AD). Our results emphasize the role of oxidative stress, particularly SOD2, in neurodegeneration triggered by environmental toxic metal mixture and give a new insight into common molecular mechanisms involved in ALS, PD and AD pathology.

Computational identification of preservatives with potential neuronal cytotoxicity

Preservatives play a vital role in cosmetics by preventing microbiological contamination for keeping products safe to use. However, a few commonly used preservatives have been suggested to be neurotoxic. Cytotoxicity to neuronal cells is commonly used as the first-tier assay for assessing chemical-induced neurotoxicity. Given the time and resources required for chemical screening, computational methods are attractive alternatives over experimental approaches in prioritizing chemicals prior to further experimental evaluations. In this study, we developed a Quantitative Structure-Activity Relationships (QSAR) model for the identification of potential neurotoxicants. A set of 681 chemicals was utilized to construct a robust prediction model using oversampling and Random Forest algorithms. Within a defined applicability domain, the independent test on 452 chemicals showed a high accuracy of 87.7%. The application of the model to 157 preservatives identified 15 chemicals potentially toxic to neuronal cells. Three of them were further validated by in vitro experiments. The results suggested that further experiments are desirable for assessing the neurotoxicity of the identified preservatives with potential neuronal cytotoxicity.

Safety assessment of drug combinations used in COVID-19 treatment: in silico toxicogenomic data-mining approach

Improvement of COVID-19 clinical condition was seen in studies where combination of antiretroviral drugs, lopinavir and ritonavir, as well as immunomodulant antimalaric, chloroquine/hydroxychloroquine together with the macrolide-type antibiotic, azithromycin, was used for patient's treatment. Although these drugs are "old", their pharmacological and toxicological profile in SARS-CoV-2 - infected patients are still unknown. Thus, by using in silico toxicogenomic data-mining approach, we aimed to assess both risks and benefits of the COVID-19 treatment with the most promising candidate drugs combinations: lopinavir/ritonavir and chloroquine/hydroxychloroquine + azithromycin. The Comparative Toxicogenomics Database (CTD; http://CTD.mdibl.org), Cytoscape software (https://cytoscape.org) and ToppGene Suite portal (https://toppgene.cchmc.org) served as a foundation in our research. Our results have demonstrated that lopinavir/ritonavir increased the expression of the genes involved in immune response and lipid metabolism (IL6, ICAM1, CCL2, TNF, APOA1, etc.). Chloroquine/hydroxychloroquine + azithromycin interacted with 6 genes (CCL2, CTSB, CXCL8, IL1B, IL6 and TNF), whereas chloroquine and azithromycin affected two additional genes (BCL2L1 and CYP3A4), which might be a reason behind a greater number of consequential diseases. In contrast to lopinavir/ritonavir, chloroquine/hydroxychloroquine + azithromycin downregulated the expression of TNF and IL6. As expected, inflammation, cardiotoxicity, and dyslipidaemias were revealed as the main risks of lopinavir/ritonavir treatment, while chloroquine/hydroxychloroquine + azithromycin therapy was additionally linked to gastrointestinal and skin diseases. According to our results, these drug combinations should be administrated with caution to patients suffering from cardiovascular problems, autoimmune diseases, or acquired and hereditary lipid disorders.

Prediction of human fetal-maternal blood concentration ratio of chemicals

Background

The measurement of human fetal-maternal blood concentration ratio (logFM) of chemicals is critical for the risk assessment of chemical-induced developmental toxicity. While a few in vitro and ex vivo experimental methods were developed for predicting logFM of chemicals, the obtained experimental results are not able to directly predict in vivo outcomes.

Methods

A total of 55 chemicals with logFM values representing in vivo fetal-maternal blood ratio were divided into training and test datasets. An interpretable linear regression model was developed along with feature selection methods. Cross-validation on training dataset and prediction on independent test dataset were conducted to validate the prediction model.

Results

This study presents the first valid quantitative structure-activity relationship model following the Organisation for Economic Co-operation and Development (OECD) guidelines based on multiple linear regression for predicting in vivo logFM values. The autocorrelation descriptor AATSC1c and information content descriptor ZMIC1 were identified as informative features for predicting logFM. After the adjustment of the applicability domain, the developed model performs well with correlation coefficients of 0.875, 0.850 and 0.847 for model fitting, leave-one-out cross-validation and independent test, respectively. The model is expected to be useful for assessing human transplacental exposure.