Evaluation of QSAR models for predicting mutagenicity: outcome of the Second Ames/QSAR international challenge project.

Quantitative structure-activity relationship (QSAR) models are powerful in silico tools for predicting the mutagenicity of unstable compounds, impurities and metabolites that are difficult to examine using the Ames test. Ideally, Ames/QSAR models for regulatory use should demonstrate high sensitivity, low false-negative rate and wide coverage of chemical space. To promote superior model development, the Division of Genetics and Mutagenesis, National Institute of Health Sciences, Japan (DGM/NIHS), conducted the Second Ames/QSAR International Challenge Project (2020-2022) as a successor to the First Project (2014-2017), with 21 teams from 11 countries participating. The DGM/NIHS provided a curated training dataset of approximately 12,000 chemicals and a trial dataset of approximately 1,600 chemicals, and each participating team predicted the Ames mutagenicity of each trial chemical using various Ames/QSAR models. The DGM/NIHS then provided the Ames test results for trial chemicals to assist in model improvement. Although overall model performance on the Second Project was not superior to that on the First, models from the eight teams participating in both projects achieved higher sensitivity than models from teams participating in only the Second Project. Thus, these evaluations have facilitated the development of QSAR models.

Malaria in long-term travelers: Infection risks and adherence to preventive measures - A prospective cohort study.

BACKGROUND: Chemoprophylaxis and anti-mosquito measures are key to preventing malaria in travelers. Long-term travelers are at higher risk than short-term travelers, but their adherence to preventive measures is lower. Our aim was to determine malaria exposure risks and predictors for adherence to malaria-preventive measures in long-term travelers. METHODS: Long-term travelers (>12 weeks) completed a weekly questionnaire about preventive measures, symptoms, and malaria treatment abroad. Blood samples were tested for seroconversion to Plasmodium falciparum anti-circumsporozoite (PfCSP) antibody. Adherence to preventive measures was defined as number of weeks of their usage divided by number of weeks in malaria-endemic areas. RESULTS: Of 561 travelers, the median travel time was 20 weeks (IQR 16-25). Eighteen were treated for malaria, all in sub-Saharan Africa. Sixteen PfCSP seroconversions were found, of whom only 3 had traveled to high-endemic areas. Of the 18 travelers treated for malaria, only one seroconverted. No associations were found between covariates and seroconversion. Neither treatment abroad nor seroconversion were reliable predictors for exposure. 'Full adherence' to chemoprophylaxis was reported by 52% (218/417) and was associated with travel to Africa, use of mefloquine, lack of prior travel history, shorter duration of travel, and use of DEET. CONCLUSIONS: The risk of malaria in this long-term travelers cohort was low. Our data confirm that anti-PfCSP seroconversion is not a reliable method to retrospectively identify incident infection, or probably exposure. Prevention efforts should focus on more experienced travellers and longer travel duration, for whom mefloquine should be considered as the first-choice chemoprophylaxis.

Could deep learning in neural networks improve the QSAR models?

Assessing chemical toxicity is a multidisciplinary process, traditionally involving in vivo, in vitro and in silico tests. Currently, toxicological goal is to reduce new tests on chemicals, exploiting all information yet available. Recent advancements in machine learning and deep neural networks allow computers to automatically mine patterns and learn from data. This technology, applied to (Q)SAR model development, leads to discover by learning the structural-chemical-biological relationships and the emergent properties. Starting from Toxception, a deep neural network predicting activity from the chemical graph image, we designed SmilesNet, a recurrent neural network taking SMILES as the only input. We then integrated the two networks into C-Tox network to make the final classification. Results of our networks, trained on a ~20K molecule dataset with Ames test experimental values, match or even outperform the current state of the art. We also extract knowledge from the networks and compare it with the available mutagenic structural alerts. The advantage over traditional QSAR modelling is that our models automatically extract the features without using descriptors. Nevertheless, the model is successful if large numbers of examples are provided and computation is more complex than in classical methods.

A large comparison of integrated SAR/QSAR models of the Ames test for mutagenicity$.

Results from the Ames test are the first outcome considered to assess the possible mutagenicity of substances. Many QSAR models and structural alerts are available to predict this endpoint. From a regulatory point of view, the recommendation from international authorities is to consider the predictions of more than one model and to combine results in order to develop conclusions about the mutagenicity risk posed by chemicals. However, the results of those models are often conflicting, and the existing inconsistency in the predictions requires intelligent strategies to integrate them. In our study, we evaluated different strategies for combining results of models for Ames mutagenicity, starting from a set of 10 diverse individual models, each built on a dataset of around 6000 compounds. The novelty of our study is that we collected a much larger set of about 18,000 compounds and used the new data to build a family of integrated models. These integrations used probabilistic approaches, decision theory, machine learning, and voting strategies in the integration scheme. Results are discussed considering balanced or conservative perspectives, regarding the possible uses for different purposes, including screening of large collection of substances for prioritization.

Results of a round-robin exercise on read-across.

A round-robin exercise was conducted within the CALEIDOS LIFE project. The participants were invited to assess the hazard posed by a substance, applying in silico methods and read-across approaches. The exercise was based on three endpoints: mutagenicity, bioconcentration factor and fish acute toxicity. Nine chemicals were assigned for each endpoint and the participants were invited to complete a specific questionnaire communicating their conclusions. The interesting aspect of this exercise is the justification behind the answers more than the final prediction in itself. Which tools were used? How did the approach selected affect the final answer?

Integrating QSAR and read-across for environmental assessment.

Read-across and QSAR have different traditions and drawbacks. We address here two main questions: (1) How do we solve the issue of the subjectivity in the evaluation of data and results, which may be particularly critical for read-across, but may have a role also for the QSAR assessment? (2) How do we take advantage of the results of both approaches to support each other? The QSAR model starts from the training set. The presence of similar chemicals with property values close to that predicted can support the result. The approach in read-across is the opposite. The assessment is focused on the few substances similar to the target. The data quality of the similar chemicals is fundamental. A risk is poor standardization in the definition of 'similarity', because different approaches may be applied. Inspired by the principles of high transparency and reproducibility, a new program for read-across, called ToxRead, has been developed and made freely available ( www.toxgate.eu ). The output of ToxRead can be compared and integrated with the output of QSAR, within a weight-of-evidence strategy. We discuss the evaluation and integration of ToxRead and QSAR with examples of the assessment of bioconcentration factors of chemicals.

ToxRead: a tool to assist in read across and its use to assess mutagenicity of chemicals.

Life sciences, and toxicology in particular, are heavily impacted by the development of methods for data collection and data analysis; they are moving from an analytical approach to a modelling approach. The scarce availability of experimental data is a known bottleneck in assessing the properties of new chemicals. Even when a model is available, the resulting predictions have to be assessed by close scrutiny of the chemicals and the biological properties of the compounds concerned. To avoid unnecessary testing, a read across strategy is often suggested and used. In this paper we discuss how to improve and standardize read across activity using ad hoc visualization and data search methods which use similarity measures and fragment search to organize in a chart a picture of all the relevant information that the expert needs to make an assessment. We show in particular how to apply our system to the case of mutagenicity.

Automatic knowledge extraction from chemical structures: the case of mutagenicity prediction.

This work proposes a new structure-activity relationship (SAR) approach to mine molecular fragments that act as structural alerts for biological activity. The entire process is designed to fit with human reasoning, not only to make the predictions more reliable but also to permit clear control by the user in order to meet customized requirements. This approach has been tested on the mutagenicity endpoint, showing marked prediction skills and, more interestingly, bringing to the surface much of the knowledge already collected in the literature as well as new evidence.

CORAL: Monte Carlo Method as a Tool for the Prediction of the Bioconcentration Factor of Industrial Pollutants.

The CORAL software (http://www.insilico.eu/coral/) has been evaluated for application in QSAR modeling of the bioconcentration factor in fish (logBCF). The data used include 237 organic substances (industrial pollutants). Six random splits of the data into sub-training (30-50 %), calibration (20-30 %), test (13-30 %), and validation sets (7-25 %) have been carried out. The following numbers display the average statistical characteristics of the models for the external validation set: correlation coefficient r(2) =0.880±0.017 and standard error of estimation s=0.559±0.131. The best models were obtained with a combined representation of the molecular structure by SMILES together with hydrogen suppressed graph.

Increased seroprevalence of IgG-class antibodies against cytomegalovirus, parvovirus B19, and varicella-zoster virus in women working in child day care.

BACKGROUND: Primary maternal infection with cytomegalovirus (CMV), parvovirus B19 (B19V), and varicella-zoster virus (VZV) may result in adverse pregnancy outcomes like congenital infection or foetal loss. Women working in child day care have an increased exposure to CMV, B19V, and VZV. By comparing the seroprevalence of IgG-class antibodies against CMV, VZV and B19V in female day care workers (DCW) with the seroprevalence in women not working in day care this study aimed to assess the association between occupation and infection. METHODS: A cross-sectional design was used. Out of a random sample of 266 day care centres, demographic data, data on work history, and blood samples were collected from 285 women from 38 centres. In addition, blood samples and basic demographics from women who participated in a cross-sectional survey of the Amsterdam population (2004) were used. All blood samples were tested for IgG-class antibodies against CMV, B19V, and VZV. RESULTS: Twenty-seven percent of the DCW were still susceptible to B19V or CMV. Working in day care was independently associated with B19V infection in all DCW (prevalence ratio [PR] 1.2; 95 % CI 1.1-1.3), and with CMV infection in DCW of European origin only (PR 1.7; 95 % CI 1.3-2.3). Almost all women born outside Europe tested seropositive for CMV (96 %). All DCW tested seropositive for VZV, compared to only 94 % of the women not working in day care. CONCLUSION: This study confirms the clear association between employment in child day care centres and infection with CMV and B19V. Intervention policies, like screening of new employees and awareness campaigns emphasizing hygienic measures among DCW, should be implemented urgently to improve the maternal health of these women and the health of their offspring.

Seroprevalence of varicella-zoster virus and predictors for seronegativity in the Amsterdam adult population.

BACKGROUND: In the Netherlands, infection with varicella-zoster virus (VZV) is considered a benign common childhood illness and routine vaccination against VZV is not done. In 1995 it was estimated that 98-100% of the adult Dutch general population is immune, yet the estimate is based on a database in which a relative small number of people of non-Dutch ethnic origin were represented. As the city of Amsterdam has large immigrant communities originating from various subtropical and tropical countries, such as Morocco, Surinam, and Turkey with probably lower VZV transmission, this study aimed to estimate the seroprevalence of VZV IgG antibodies (anti-VZV) among various ethnic groups in Amsterdam, and identify factors associated with seronegative VZV status. METHODS: The study was a cross-sectional survey of the Amsterdam population (2004), and the study sample was stratified by age and ethnicity, with deliberate oversampling of minority ethnic groups. Serum samples obtained from 1,341 residents in 2004 were tested for antibodies to VZV. Basic demographic data (gender, age, country of birth, age at immigration and number of children) were also available. RESULTS: The anti-VZV seroprevalence in the overall Amsterdam population was estimated to be 94% (95% confidence intervals; 92-96%). Regarding ethnic origin, first generation immigrants (Moroccan immigrants 90%, Surinamese or Antillean immigrants 91%, and Turkish 92%), especially those that migrated after the age of 11 years, were more likely to be anti-VZV seronegative compared to those arriving at an earlier age or those born in the Netherlands (97-98%). Both ethnic origin and generation of immigration were positive predictors for IgG seronegativity to VZV (p<0.015). No other predictors for seronegativity were found. CONCLUSION: The results of this study imply that about 4-8% of the general adult Amsterdam population is still susceptible to infection with VZV, and that susceptibility is even higher in some immigrant groups. When assessing the risk of infection after VZV exposure alertness is needed for vulnerable persons like pregnant women, patients with hematological malignancies or organ transplants in particular among first-generation immigrants.

CORAL: QSPR modeling of rate constants of reactions between organic aromatic pollutants and hydroxyl radical.

The rate constants (K(OH)) of reactions between 78 organic aromatic pollutants and hydroxyl radical were examined. Simplified molecular input line entry system was used as representation of the molecular structure of the pollutants. Quantitative structure-property relationships was developed using CORAL software (http://www.insilico.eu/CORAL) for four random splits of the data into the subtraining, calibration, and test sets. The obtained results reveal good predictive potential of the applied approach: correlation coefficients (r(2)) for the test sets of the four random splits are 0.75, 0.91, 0.84, and 0.80. Using the Monte Carlo method CORAL software generated the optimal descriptors for one-variable models. The reproducibility of each model was tested performing three runs of the Monte Carlo optimization. The current data were compared to previous results and discussed.

Low risk of hepatitis E among Dutch short-term travelers.

Hepatitis E is endemic in (sub)tropical countries while only sporadic cases have been described in industrialized countries. In a prospective study among 1270 short-term Dutch travelers to (sub)tropical countries we found no seroconversion to anti-hepatitis E virus (HEV) antibodies, indicating a very low risk for travelers to acquire a hepatitis E infection.

CORAL: QSAR models for acute toxicity in fathead minnow (Pimephales promelas).

CORrelation And Logic (CORAL) is a software that generates quantitative structure activity relationships (QSAR) for different endpoints. This study is dedicated to the QSAR analysis of acute toxicity in Fathead minnow (Pimephales promelas). Statistical quality for the external test set is a complex function of the split (into training and test subsets), the number of epochs of the Monte Carlo optimization, and the threshold that is a criterion for dividing the correlation weights into two classes rare (blocked) and not rare (active). Computational experiments with three random splits (data on 568 compounds) indicated that this approach can satisfactorily predict the desired endpoint (the negative decimal logarithm of the 50% lethal concentration, in mmol/L, pLC50). The average correlation coefficients (r2) are 0.675 ± 0.0053, 0.824 ± 0.0242, 0.787 ± 0.0101 for subtraining, calibration, and test set, respectively. The average standard errors of estimation (s) are 0.837 ± 0.021, 0.555 ± 0.047, 0.606 ± 0.049 for subtraining, calibration, and test set, respectively. The CORAL software together with three random splits into subtraining, calibration, and test sets can be downloaded on the Internet (http://www.insilico.eu/coral/).

CORAL: classification model for predictions of anti-sarcoma activity.

A modified version of the CORAL software (http://www.insilico.eu/coral) allows building up the classification model for the case of the Yes/No data on the anti-sarcoma activity of organic compounds. Three random splits into the sub-training, calibration, and test sets of the data for 3017 compounds were examined. The performance of the proposed approach is satisfactory. The average values of the statistical characteristics for external test set on three random splits are as follows: n=1173-1234, sensitivity = 0.8903±0.0390, specificity = 0.9869±0.0013, and accuracy = 0.9759±0.0043. Mechanistic interpretation of the suggested model is discussed.

CORAL: quantitative structure-activity relationship models for estimating toxicity of organic compounds in rats.

For six random splits, one-variable models of rat toxicity (minus decimal logarithm of the 50% lethal dose [pLD50], oral exposure) have been calculated with CORAL software (http://www.insilico.eu/coral/). The total number of considered compounds is 689. New additional global attributes of the simplified molecular input line entry system (SMILES) have been examined for improvement of the optimal SMILES-based descriptors. These global SMILES attributes are representing the presence of some chemical elements and different kinds of chemical bonds (double, triple, and stereochemical). The "classic" scheme of building up quantitative structure-property/activity relationships and the balance of correlations (BC) with the ideal slopes were compared. For all six random splits, best prediction takes place if the aforementioned BC along with the global SMILES attributes are included in the modeling process. The average statistical characteristics for the external test set are the following: n = 119 ± 6.4, R(2) = 0.7371 ± 0.013, and root mean square error = 0.360 ± 0.037.

CORAL: building up the model for bioconcentration factor and defining it's applicability domain.

CORAL (CORrelation And Logic) software can be used to build up the quantitative structure--property/activity relationships (QSPR/QSAR) with optimal descriptors calculated with the simplified molecular input line entry system (SMILES). We used CORAL to evaluate the applicability domain of the QSAR models, taking a model of bioconcentration factor (logBCF) as example. This model's based on a large training set of more than 1000 chemicals. To improve the model is predictivity and reliability on new compounds, we introduced a new function, which uses the Delta(obs) = logBCF(expr)--logBCF(calc) of the predictions on the chemicals in the training set. With this approach, outliers are eliminated from the phase of training. This proved useful and increased the model's predictivity.

Declining incidence of imported malaria in the Netherlands, 2000-2007.

BACKGROUND: To describe the epidemiology and trends of imported malaria in the Netherlands from 2000 through 2007. METHODS: Based on national surveillance data regarding all reported infections of imported malaria, diagnosed 2000 through 2007, incidence and trends of imported malaria in the Netherlands were estimated. Travellers statistics were used to estimate incidence, and data on malaria chemoprophylaxis prescriptions were used to estimate the number of unprotected travellers. RESULTS: Importation of malaria to the Netherlands is declining even as more travellers visit malaria-endemic countries. On average, 82% were acquired in sub-Saharan Africa, and 75% were caused by Plasmodium falciparum. The overall incidence in imported falciparum malaria fell from 21.5 to 6.6/10,000 of unprotected travellers. The percentage of unprotected travellers rose from 47% to 52% of all travellers. The incidence of imported falciparum infections is greatest from Middle and West Africa, and decreased from 121.3 to 36.5/10,000 travellers. The import of malaria from this region by immigrants visiting friends and relatives (VFR) decreased from 138 infections in 2000, to 69 infections in 2007. CONCLUSION: The annual number of imported malaria shows a continuing declining trend, even with an increasing number of travellers visiting malaria endemic countries. VFR import less malaria than previously, and contribute largely to the declining incidence seen. The decline is not readily explained by increased use of chemoprophylaxis and may reflect a reduced risk of infection due to decreasing local malaria transmission as observed in some malaria endemic areas. Nevertheless, the increasing number of unprotected travellers remains worrisome.

A new bioconcentration factor model based on SMILES and indices of presence of atoms.

Indices of the presence of atoms (IPA) encode the presence or absence of atoms, such as nitrogen, oxygen, sulphur, phosphorus, fluorine, chlorine, and bromine in a molecule. They are calculated with the simplified molecular input line entry system (SMILES). Using the Monte Carlo method for correlation weights of these indices, one can improve the predictive ability of optimal SMILES-based descriptors in quantitative structure-activity relationships (QSAR) for bioconcentration factor. The model without IPA gave the following results: n=503, r(2)=0.6803, q(2)=0.6781, s=0.759, F=1066 (subtraining set); n=322, r(2)=0.8181, r(pred)(2)=0.8159, s=0.565, F=1439 (calibration set); n=105, r(2)=0.6703, r(pred)(2)=0.6577, R(m)(2)=0.6628, s=0.728, F=209 (test set); n=106, r(2)=0.6624, r(pred)(2)=0.6502, R(m)(2)=0.6212, s=0.757, F=204 (validation set) The model with IPA gave: n=503, r(2)=0.7082, q(2)=0.7062, s=0.725, F=1216 (subtraining set); n=322, r(2)=0.8401, r(pred)(2)=0.8383, s=0.528, F=1682 (calibration set); n=105, r(2)=0.7489, r(pred)(2)=0.7402, R(m)(2)=0.7252, s=0.637, F=307 (test set); n=106, r(2)=0.7306, r(pred)(2)=0.7217, R(m)(2)=0.7010, s=0.680, F=282 (validation set).