Hybrid Class Balancing Approach for Chemical Compound Toxicity Prediction.

INTRODUCTION: Computational methods are crucial for efficient and cost-effective drug toxicity prediction. Unfortunately, the data used for prediction is often imbalanced, resulting in biased models that favor the majority class. This paper proposes an approach to apply a hybrid class balancing technique and evaluate its performance on computational models for toxicity prediction in Tox21 datasets. METHODS: The process begins by converting chemical compound data structures (SMILES strings) from various bioassay datasets into molecular descriptors that can be processed by algorithms. Subsequently, Undersampling and Oversampling techniques are applied in two different schemes on the training data. In the first scheme (Individual), only one balancing technique (Oversampling or Undersampling) is used. In the second scheme (Hybrid), the training data is divided according to a ratio (e.g., 90-10), applying a different balancing technique to each proportion. We considered eight resampling techniques (four Oversampling and four Undersampling), six molecular descriptors (based on MACCS, ECFP, and Mordred), and five classification models (KNN, MLP, RF, XGB and SVM) over 10 bioassay datasets to determine the configurations that yield the best performance. RESULTS: We defined three testing scenarios: without balancing techniques (baseline), Individual, and Hybrid. We found that using the ENN technique in the MACCS-MLP combination resulted in a 10.01% improvement in performance. The increase for ECFP6-2048 was 16.47% after incorporating a combination of the SMOTE (10%) and RUS (90%) techniques. Meanwhile, using the same combination of techniques, MORDRED-XGB showed the most significant increase in performance, achieving a 22.62% improvement. CONCLUSION: Integrating any of the class balancing schemes resulted in a minimum of 10.01% improvement in prediction performance compared to the best baseline configuration. In this study, Undersampling techniques were more appropriate due to the significant overlap among samples. By eliminating specific samples from the predominant class that are close to the minority class, this overlap is greatly reduced.

Toxicological Screening of Four Bioactive Citroflavonoids: In Vitro, In Vivo, and In Silico Approaches.

Many studies describe different pharmacological effects of flavonoids on experimental animals and humans. Nevertheless, few ones are confirming the safety of these compounds for therapeutic purposes. This study aimed to investigate the preclinical safety of naringenin, naringin, hesperidin, and quercetin by in vivo, in vitro, and in silico approaches. For this, an MTT-based cytotoxicity assay in VERO and MDCK cell lines was performed. In addition, acute toxicity was evaluated on Wistar rats by OECD Guidelines for the Testing of Chemicals (Test No. 423: Acute Oral Toxicity-Class Method). Furthermore, we used the ACD/Tox Suite to predict toxicological parameters such as hERG channel blockade, CYP450 inhibition, and acute toxicity in animals. The results showed that quercetin was slightly more cytotoxic on cell lines (IC50 of 219.44 ± 7.22 mM and 465.41 ± 7.44 mM, respectively) than the other citroflavonoids. All flavonoids exhibited an LD50 value > 2000 mg/kg, which classifies them as low-risk substances as OECD guidelines established. Similarly, predicted LD50 was LD50 > 300 to 2000 mg/kg for all flavonoids as acute toxicity assay estimated. Data suggests that all these flavonoids did not show significant toxicological effects, and they were classified as low-risk, useful substances for drug development.

Comparison between UVA- and zero-valent iron-activated persulfate processes for degrading propylparaben.

Conventional wastewater treatments are not efficient in removing parabens, which may thus end up in surface waters, posing a threat to aquatic biota and human health. As an alternative treatment, persulfate (PS)-driven advanced oxidation technologies have gained growing attention for removing these pollutants. In this study, the degradation of propylparaben (PrP) by UVA- and zero-valent iron (ZVI)-activated persulfate was investigated. The effects of initial PS concentration ([PS]0) and irradiance or ZVI concentration were explored using the Doehlert experimental design. For the UVA-activated system, the specific PrP degradation rate (k) and percent removal were consistently higher for increasing [PS]0 and irradiance, varying in the ranges 0.0053-0.0192 min-1 and 37.9-77.3%, respectively. In contrast, extremely fast PrP degradation was achieved through the ZVI/PS process (0.3304 < k < 0.9212 min-1), with removal percentages above 97.5%; in this case, paraben degradation was hindered for a ZVI dosage beyond 40 mg L-1. Regarding toxicity, ECOSAR predictions suggest that the degradation products elucidated by LC-MS/MS are less toxic than PrP toward fish, daphnid, and green algae. In addition, both processes showed to be strongly dependent on the water matrix, being ZVI/PS more impacted for a MBR effluent, although its performance was much better than that exhibited by the UVA-driven process (t1/2 of 65.4 and 276.1 min, respectively).

Human adipose-derived stem cells (ADSC) and human periodontal ligament stem cells (PDLSC) as cellular substrates of a toxicity prediction assay.

With the increasing need to develop in vitro assays to replace animal use, human stem cell-derived methods are emerging and showing outstanding contributions to the toxicological screening of substances. Adult human stem cells such as adipose-derived stem cells (ADSC) and periodontal ligament stem cells (PDLSC) were used as cell substrates for a cytotoxicity assay and toxicity prediction using the neutral red uptake (NRU) assay. First, primary cell cultures from three independent donors, from each tissue source, were characterized as mesenchymal stem cells (MSC) by plastic adherence and appropriate immunophenotype for MSC markers (positive for CD90, CD73, and CD105 and negative for CD11b, CD34, CD45, HLADR, and CD19). Furthermore, ADSC and PDLSC were able to differentiate into adipocytes and osteoblasts when maintained under the same culture conditions previously established for the NRU assay. NRU assays for three reference test substances were performed. R2 was higher than 0.85 for all conditions, showing the feasibility to calculate IC50 values. The IC50 values were then used to predict the LD50 of the test substances, which were comparable to previous results and the ICCVAM standard test report. Primary ADSC and PDLSC showed the potential to be considered as additional models for use in cytotoxicity assays.

Potential of the Biotic Ligand Model (BLM) to Predict Copper Toxicity in the White-Water of the Solimões-Amazon River.

In this study, we evaluated the capacity of the Biotic Ligand Model (BLM) to predict copper toxicity in white-waters of the Solimões-Amazon River. LC50 tests using the species Otocinclus vittatus (Regan, 1904) were performed with Solimões-Amazon river water (100%) at 20%, 40%, 60%, and 80% dilutions. A sevenfold decrease in both dissolved and total Cu toxicity was observed in the experiment conducted with 100% when compared to 20% white-water, indicating that physicochemical characteristics of white-water attenuate Cu toxicity. There was agreement between the observed LC50 and the LC50 predicted by the BLM after the adjustment of critical accumulation concentration (LA50) for O. vittatus. BLM modeling indicated that dissolved organic carbon (DOC) and pH were the most important water parameters influencing Cu toxicity, followed by Ca2+. Our results highlight the first evidence that the BLM presents potential to predict Cu toxicity to aquatic organisms in the white-water of the Solimões-Amazon River.

Sida tuberculata (Malvaceae): a study based on development of extractive system and in silico and in vitro properties

Sida tuberculata (Malvaceae) is a medicinal plant traditionally used in Brazil as an antimicrobial and anti-inflammatory agent. Here, we aimed to investigate the different extractive techniques on phytochemical parameters, as well as to evaluate the toxicity and antioxidant capacity of S. tuberculata extracts using in silico and in vitro models. Therefore, in order to determine the dry residue content and the main compound 20-hydroxyecdysone (20E) concentration, extracts from leaves and roots were prepared testing ethanol and water in different proportions. Extracts were then assessed by Artemia salina lethality test, and toxicity prediction of 20E was estimated. Antioxidant activity was performed by DPPH and ABTS radical scavenger assays, ferric reducing power assay, nitrogen derivative scavenger, deoxyribose degradation, and TBARS assays. HPLC evaluation detected 20E as main compound in leaves and roots. Percolation method showed the highest concentrations of 20E (0.134 and 0.096 mg/mL of extract for leaves and roots, respectively). All crude extracts presented low toxic potential on A. salina (LD50 >1000 µg/mL). The computational evaluation of 20E showed a low toxicity prediction. For in vitro antioxidant tests, hydroethanolic extracts of leaves were most effective compared to roots. In addition, hydroethanolic extracts presented a higher IC50 antioxidant than aqueous extracts. TBARS formation was prevented by leaves hydroethanolic extract from 0.015 and 0.03 mg/mL and for roots from 0.03 and 0.3 mg/mL on egg yolk and rat tissue, respectively (P<0.05). These findings suggest that S. tuberculata extracts are a considerable source of ecdysteroids and possesses a significant antioxidant property with low toxic potential.

The use of human adipose-derived stem cells based cytotoxicity assay for acute toxicity test.

Human adipose-derived stem cells (ADSC) were evaluated as cell culture model for cytotoxicity assay and toxicity prediction by using the neutral red uptake assay (NRU). In this study, we compared ADSC and the murine cell line BALB/c 3T3 clone A31 to predict the toxicity of 12 reference substances as recommended by the Interagency Coordinating Committee on the Validation of Alternative Methods. We predicted the LD50 for RC-rat-only weight and RC-rat-only millimole regressions for both cell culture models. For RC rat-only weight regression, both cells had the same accordance (50%), while for RC rat-only millimole regression, the accordance was 50% for ADSC and 42% for 3T3s. Thus, ADSC have similar capability for GHS class prediction as the 3T3 cell line for the evaluated reference substances. Therefore, ADSCs showed the potential to be considered a novel model for use in evaluating cytotoxicity in drug development and industry as well as for regulatory purposes to reduce or replace the use of laboratory animals with acceptable sensitivity for toxicity prediction in humans. These cells can be used to complete the results from other models, mainly because of its human origin. Moreover, it is less expensive in comparison with other existing models.