Molecular Generation, QSAR, and Molecular Dynamic Simulation Studies for Virtual Screening of DNA Polymerase Theta Inhibitors.

AIMS: The machine learning-based QSAR modeling procedure, molecular generations, and molecular dynamic simulations were applied to virtually screen the DNA polymerase theta inhibitors. BACKGROUND: The DNA polymerase theta (Polθ or POLQ) is an attractive target for treatments of homologous recombination deficient (such as BRCA deficient) cancers. There are no approved drugs for targeting POLQ, and only one inhibitor is in Phase Ⅱclinical trials; thus, it is necessary to develop novel POLQ inhibitors. OBJECTIVES: To build machine learning models that predict the bioactivities of POLQ inhibitors. To build molecular generation models that generate diverse molecules. To virtually screen the generated molecules by the machine learning models. To analyze the binding modes of the screening results by molecular dynamic simulations. METHODS: In the present work, 325 inhibitors with POLQ polymerase domain bioactivities were Collected. Two machine learning methods, random forest and deep neural network, were used for building the ligand- and structure-based quantitative structure-activity relationship (QSAR) models. The substructure replacement-based method and transfer learning-based deep recurrent neural network method were used for molecular generations. Molecular docking and consensus QSAR models were carried out for virtual screening. The molecular dynamic simulations and MM/GBSA binding free energy calculation and decomposition were used to further analyze the screening results. RESULTS: The MCC values of the best ligand- and structure-based consensus QSAR models reached 0.651 and 0.361 for the test set, respectively. The machine learning-based docking scores had better-predicted ability to distinguish the highly and weakly active poses than the original docking scores. The 96490 molecules were generated by both molecular generation methods, and 10 molecules were retained by virtual screening. Four favorable interactions were concluded by molecular dynamic simulations. CONCLUSION: We hope that the screening results and the binding modes are helpful for designing the highly active POLQ polymerase inhibitors and the models of the molecular design workflow can be used as reliable tools for drug design.

Electrocardiogram identification based on data generative network and non-fiducial data processing.

Nowadays, the use of biological signals as a criterion for identity recognition has gained increasing attention from various organizations and companies. Therefore, it has become crucial to have a biometric identity recognition method that is fast and accurate. In this paper, we propose a linear electrocardiogram (ECG) data preprocessing algorithm based on Kalman filters for rapid noise data filtering (wavelet transform filtering algorithm). Additionally, we introduce a generative network model called Data Generation Strategy Network (DRCN) based on generative networks. The DRCN is employed to augment training samples for convolutional classification networks, ultimately improving the classification performance of the model. Through the final experiments, our method successfully reduced the average misidentification rate of ECG-based identity recognition to 2.5%, and achieved an average recognition rate of 98.7% for each category, significantly surpassing previous achievements. In the future, this method is expected to be widely applied in the field of ECG-based identity recognition.

Association between BMI trajectories in late-middle age and subsequent dementia risk in older age: a 26-year population-based cohort study.

BACKGROUND: The association between body mass index (BMI) and dementia risk differs depending on follow-up time and age at BMI measurement. The relationship between BMI trajectories in late-middle age (50-65 years old) and the risk of dementia in older age (> 65 years old) has not been revealed. METHODS: In the present study, participants from the Health and Retirement Study were included. BMI trajectories were constructed by combining BMI trend and variation information. The association between BMI trajectories at the age of 50-65 years and dementia risk after the age of 65 years was investigated. Participants with European ancestry and information on polygenic scores for cognitive performance were pooled to examine whether genetic predisposition could modify the association. RESULTS: A total of 10,847 participants were included in the main analyses. A declining BMI trend and high variation in late-middle age were associated with the highest subsequent dementia risk in older age compared with an ascending BMI trend and low variation (RR = 1.76, 95% CI = 1.45-2.13). Specifically, in stratified analyses on BMI trajectories and dementia risk based on each individual's mean BMI, the strongest association between a declining BMI trend with high variation and elevated dementia risk was observed in normal BMI group (RR = 2.66, 95% CI = 1.72-4.1). Similar associations were found when participants were stratified by their genetic performance for cognition function without interaction. CONCLUSIONS: A declining BMI trend and high variation in late-middle age were associated with a higher risk of dementia. Early monitoring of these individuals is needed to prevent dementia in older individuals.

Liquid‒liquid phase separation: roles and implications in future cancer treatment.

Liquid‒liquid phase separation (LLPS) is a phenomenon driven by weak interactions between biomolecules, such as proteins and nucleic acids, that leads to the formation of distinct liquid-like condensates. Through LLPS, membraneless condensates are formed, selectively concentrating specific proteins while excluding other molecules to maintain normal cellular functions. Emerging evidence shows that cancer-related mutations cause aberrant condensate assembly, resulting in disrupted signal transduction, impaired DNA repair, and abnormal chromatin organization and eventually contributing to tumorigenesis. The objective of this review is to summarize recent advancements in understanding the potential implications of LLPS in the contexts of cancer progression and therapeutic interventions. By interfering with LLPS, it may be possible to restore normal cellular processes and inhibit tumor progression. The underlying mechanisms and potential drug targets associated with LLPS in cancer are discussed, shedding light on promising opportunities for novel therapeutic interventions.

Evaluating the Causal Association between Inflammatory Bowel Disease and Risk of Atherosclerotic Cardiovascular Disease: Univariable and Multivariable Mendelian Randomization Study.

BACKGROUND: Observational studies suggested that inflammatory bowel disease (IBD) (i.e., Crohn's disease [CD] and ulcerative colitis [UC]) is associated with an increased risk of atherosclerotic cardiovascular disease (ASCVD), including coronary artery disease (CAD) and ischemic stroke. However, it is still unclear whether the observed associations causally exist. Thus, we aim to examine the potential effect of IBD, CD, and UC on the risk of CAD and ischemic stroke, using a two-sample Mendelian randomization (MR) study. METHODS: Genetic instruments for IBD, CD, and UC were retrieved from the latest published genome-wide association studies (GWASs) of European ancestry. GWAS summary data for instrument-outcome associations were gathered from four independent resources: CARDIoGRAMplusC4D Consortium, MEGASTROKE consortium, FinnGen, and UK Biobank. The inverse variance weighted (IVW) method and multiple pleiotropy-robust approaches were conducted and, subsequently, combined in a fixed-effect meta-analysis. Moreover, multivariable MR (MVMR) analysis was conducted to adjust for potential influencing instrumental variables. RESULTS: The IVW method revealed no causal effect of IBD on the risk of CAD (overall IBD on CAD: OR 1.003, 95%CI 0.982 to 1.025; CD on CAD: OR 0.997, 95%CI 0.978 to 1.016; UC on CAD: OR 0.986, 95%CI 0.963 to 1.010) or the risk of ischemic stroke (overall IBD on ischemic stroke: OR 0.994, 95%CI 0.970 to 1.018; CD on ischemic stroke: OR 0.996, 95%CI 0.979 to 1.014; UC on ischemic stroke: OR 0.999, 95%CI 0.978 to 1.020). The results of the meta-analysis and MVMR remained consistent. CONCLUSION: Our MR analysis does not support a causal effect of IBD on CAD and ischemic stroke, and previous results from observational studies might be biased through uncontrolled confoundings (such as IBD-specific medications and detection bias, etc.) that warrant further research.

Superscattering of water waves.

Inspired by the concept of superscattering in optics, we for the first time theoretically predict and experimentally demonstrate the superscattering phenomenon in water waves. The subwavelength superscatterer is constructed by multi-layered concentric cylinders with an inhomogeneous depth profile. The superscatterer breaks the long-held single-channel scattering limit by several times and thus significantly enhances the total scattering strength. The underlying mechanism originates from the near degeneracy of the resonances of multiple channels. We fabricate the superscatterer prototype and experimentally measure the near-field patterns, which are consistent with theoretical prediction and numerical simulation. Our study opens a new avenue to strengthen water-wave scattering and deepen the understanding in water waves, which can be useful for ocean energy harvesting and harbor protection.

Function of stem cells in radiation-induced damage.

PURPOSE: The aim of this review is to discuss previous studies on the function of stem cells in radiation-induced damage, and the factors affecting these processes, in the hope of improving our understanding of the different stem cells and the communication networks surrounding them. This is essential for the development of effective stem cell-based therapies to regenerate or replace normal tissues damaged by radiation. CONCLUSION: In salivary glands, senescence-associated cytokines and inflammation-associated cells have a greater effect on stem cells. In the intestinal glands, Paneth cells strongly affect stem cell-mediated tissue regeneration after radiation treatment. In the pancreas, ß-cells as well as protein C receptor positive (Procr) cells are expected to be key cells in the treatment of diabetes. In the bone marrow, a variety of cytokines such as CXC-chemokine ligand 12 (CXCL12) and stem cell factor (SCF), contribute to the functional recovery of hematopoietic stem cells after irradiation.

A "3S+f" Nephrometry Score System to Predict the Clinical Outcomes of Laparoscopic Nephron-Sparing Surgery.

Background: When we treat renal cell carcinoma by laparoscopic nephron-sparing surgery (NSS), it is essential to use an evaluation system to predict clinical outcomes. Hitherto, there are more than a dozen nephrometry score systems. In this study, through assessing the correlations between nephrometry score systems and clinical outcomes, we aim to provide a novel nephrometry score system-the "3S+f" score system-to simplify the evaluation of technical complexity of partial nephrectomy. Methods: We retrospectively collected the data of 131 patients who underwent NSS, which was performed by a single surgeon (SZ) from January 2013 to July 2018 at Peking University Third Hospital. The "3S+f" score system contains four parameters: "size, side, site, and fat", all of which can be obtained from preoperative imaging data. We evaluated the correlations between the "3S+f" score and clinical outcomes, and compared R.E.N.A.L. score and PADUA score. Results: All the three nephrometry score systems were related to some clinical outcomes in univariate analyses. In multivariate regression models, the "3S+f" score, the R.E.N.A.L. score, and the PADUA score were significantly associated with operative time (p = 0.016, p = 0.035, and p = 0.001, respectively) and warm ischemia time (all p = 0.008, p < 0.001, and p < 0.001, respectively). "3S+f" was also significantly related to extubation time > 5 days (p = 0.018). In predicting operative time > 120 min and extubation time >5 days from ROC curves, the AUCs of the "3S+f" score (0.717 and 0.652, respectively) were larger than both the R.E.N.A.L (0.598 and 0.554, respectively) and PADUA (0.600 and 0.542, respectively) score systems. Conclusion: A novel nephrometry score system-the "3S+f" score system-shows equivalent correlation and the ability in predicting clinical outcomes when compared to the R.E.N.A.L. score system and the PADUA score system, which can describe renal tumors.

Indirect bioleaching recovery of valuable metals from electroplating sludge and optimization of various parameters using response surface methodology (RSM).

Electroplating sludge contains amounts of valuable/toxic metals as a typical hazardous solid waste, but existing technology is hard to simultaneously gain the high recovery of valuable metals and its convert into general solid waste. In this study, indirect bioleaching process was optimized by using RSM for high recovery of four valuable metals (Ni, Cu, Zn and Cr) from electroplating sludge and its shift into general waste. The results showed that the maximum leaching rate respectively was 100% for Ni, 96.5% for Cu, 100% for Zn and 76.1% for Cr at the optimal conditions. In particular, bioleaching saw a much better performance than H2SO4 leaching in removal of highly toxic Cr (76.1% vs. 30.2%). The extraction efficiency of Cr by H2SO4 leaching sharply rose to 72.6% in the presence of 9.0 g/L Fe3+, suggesting that Fe3+ played an important role in the bioleaching of Cr. Based on bioleaching dynamics analysis, it was speculated that Fe3+ passes through the solid shell and enter inside the sludge to attack Cr assisting by extracellular polymeric substances (EPS), leading to high extraction and low residue of Cr. Meanwhile, due to high-efficient release and removal of valuable/toxic metals by bioleaching, the bioleached residues successfully degraded into general based on TCLP test and can be reused as construction material safely.

Building 2D classification models and 3D CoMSIA models on small-molecule inhibitors of both wild-type and T790M/L858R double-mutant EGFR.

Epidermal growth factor receptor (EGFR) has received widespread attention because it is an important target for anticancer drug design. Mutations in the EGFR, especially the T790M/L858R double mutation, have made cancer treatment more difficult. We herein built the structure-activity relationship models of small-molecule inhibitors on wild-type and T790M/L858R double-mutant EGFR with a whole dataset of 379 compounds. For 2D classification models, we used ECFP4 fingerprints to build support vector machine and random forest models and used SMILES to build self-attention recurrent neural network models. Each of all six models resulted in an accuracy of above 0.87 and the Matthews correlation coefficient value of above 0.76 on the test set, respectively. We concluded that inhibitors containing anilinoquinoline and methoxy or fluoro phenyl are highly active against wild EGFR. Substructures such as anilinopyrimidine, acrylamide, amino phenyl, methoxy phenyl, and thienopyrimidinyl amide appeared more in highly active inhibitors against double-mutant EGFR. We also used self-organizing map to cluster the inhibitors into six subsets based on ECFP4 fingerprints and analyzed the activity characteristics of different scaffolds in each subset. Among them, three datasets, which are based on pteridin, anilinopyrimidine, and anilinoquinoline scaffold, were selected to build 3D comparative molecular similarity analysis models individually. Models with the leave-one-out coefficient of determination (q2) above 0.65 were selected, and five descriptor types (steric, electrostatic, hydrophobic, donor, and acceptor) were used to study the effects of side chains of inhibitors on the activity against wild-type and mutant-type EGFR.

Discovery of Novel Epidermal Growth Factor Receptor (EGFR) Inhibitors Using Computational Approaches.

The epidermal growth factor receptor (EGFR) signaling pathway plays an important role in cell growth, proliferation, differentiation, and other physiological processes, which makes the EGFR a promising target for anticancer therapies. The discovery of novel EGFR inhibitors may provide a solution to the problem of drug resistance. In this work, we performed a ligand-based virtual screening (LBVS) protocol for finding novel EGFR inhibitors from a 5.3 million compound library. First, the 3D shape-based similarity was used to obtain structurally novel EGFR inhibitors. In this study, we tried three queries; two were crystal structures and one was generated from deep generative models of graphs (DGMG). Next, we have built four structure-activity relationship (SAR) models and three quantitative structure-activity relationship (QSAR) models based on an SVM method for further screening of highly active EGFR inhibitors. Experimental validations led to the identification of nine hits out of 18 tested compounds. Among them, hit 1, hit 5, and hit 6 had IC50 values around 80 nM against EGFR whose interactions with EGFR were further investigated by molecular dynamics simulations.

Classification models and SAR analysis on CysLT1 receptor antagonists using machine learning algorithms.

Cysteinyl leukotrienes 1 (CysLT1) receptor is a promising drug target for rhinitis or other allergic diseases. In our study, we built classification models to predict bioactivities of CysLT1 receptor antagonists. We built a dataset with 503 CysLT1 receptor antagonists which were divided into two groups: highly active molecules (IC50 < 1000 nM) and weakly active molecules (IC50 ≥ 1000 nM). The molecules were characterized by several descriptors including CORINA descriptors, MACCS fingerprints, Morgan fingerprint and molecular SMILES. For CORINA descriptors and two types of fingerprints, we used the random forests (RF) and deep neural networks (DNN) to build models. For molecular SMILES, we used recurrent neural networks (RNN) with the self-attention to build models. The accuracies of test sets for all models reached 85%, and the accuracy of the best model (Model 2C) was 93%. In addition, we made structure-activity relationship (SAR) analyses on CysLT1 receptor antagonists, which were based on the output from the random forest models and RNN model. It was found that highly active antagonists usually contained the common substructures such as tetrazoles, indoles and quinolines. These substructures may improve the bioactivity of the CysLT1 receptor antagonists.

Predictive value of preoperative lymphocyte-to-monocyte ratio on survival outcomes in bladder cancer patients after radical cystectomy.

Purpose: To determine the prognostic significance of the pre-operative lymphocyte-to-monocyte (LMR) in patients with bladder cancer (BCa) who underwent radical cystectomy (RC), and to assess its prognostic benefit compared to models relying solely on clinicopathological factors. Materials and Methods: A retrospective analysis of the 342 BCa patients undergoing RC at our institution from 2004 to 2017 was conducted to assess LMR prognostic significance. Overall survival (OS) and cancer-specific survival (CSS) were assessed using the Kaplan-Meier method. Cox regression models identified risk factors for survival outcomes. Two new models were developed based on basal models to predict OS and CSS at 1, 3, and 5 years after RC. The accuracy of the new models was evaluated using receiver operating characteristic curves as well as the concordance index. We also conducted decision curve analysis (DCA) to assess their net benefit. Results: An association between excellent long-term patient survival outcomes and higher LMR levels was observed. The median OS and CSS for higher LMR level in patients was 98.8 months and >120 months, respectively. Cox regression multivariate analysis showed that pre-operative LMR, as a continuous variable, was an independent survival outcome predictor (p<0.001). The utilization of LMR in the standard model resulted in significant discriminatory improvement in OS (5.6%, p<0.001) and CSS (4.9%, p=0.001) prediction. Moreover, as shown in DCA, utilization of the new model, including LMR, resulted in a net benefit compared to base models for predicting OS and CSS at 1, 3, and 5 years. Conclusions: An independent association was observed between higher pre-operative LMR in BCa patients undergoing RC and significantly better OS and CSS. In addition, a significant improvement in predictive accuracy was observed with LMR inclusion in multiparametric prediction tools. Therefore, LMR may be utilized in pre-operative patient risk stratification to assist in patient counseling and clinical decision making.

Fingerprint-based computational models of 5-lipo-oxygenase activating protein inhibitors: Activity prediction and structure clustering.

Inflammatory diseases can be treated by inhibiting 5-lipo-oxygenase activating protein (FLAP). In this study, a data set containing 2,112 FLAP inhibitors was collected. A total of 25 classification models were built by five machine learning algorithms with five different types of fingerprints. The best model, which was built by support vector machine algorithm with ECFP_4 fingerprint had an accuracy and a Matthews correlation coefficient of 0.862 and 0.722 on the test set, respectively. The predicted results were further evaluated by the application domain dSTD-PRO (a distance between one compound to models). Each compound had a dSTD-PRO value, which was calculated by the predicted probabilities obtained from all 25 models. The application domain results suggested that the reliability of predicted results depended mainly on the compounds themselves rather than algorithms or fingerprints. A group of customized 10-bit fingerprint was manually defined for clustering the molecular structures of 2,112 FLAP inhibitors into eight subsets by K-Means. According to the clustering results, most of inhibitors in two subsets (subsets 2 and 4) were highly active inhibitors. We found that aryl oxadiazole/oxazole alkanes, biaryl amino-heteroarenes, two aromatic rings (often N-containing) linked by a cyclobutene group, and 1,2,4-triazole group were typical fragments in highly active inhibitors.

Quantitative Structure-Activity Relationship Study for HIV-1 LEDGF/p75 Inhibitors.

BACKGROUND: HIV-1 Integrase (IN) is an important target for the development of the new anti-AIDS drugs. HIV-1 LEDGF/p75 inhibitors, which block the integrase and LEDGF/p75 interaction, have been validated for reduction in HIV-1 viral replicative capacity. METHODS: In this work, computational Quantitative Structure-Activity Relationship (QSAR) models were developed for predicting the bioactivity of HIV-1 integrase LEDGF/p75 inhibitors. We collected 190 inhibitors and their bioactivities in this study and divided the inhibitors into nine scaffolds by the method of T-distributed Stochastic Neighbor Embedding (TSNE). These 190 inhibitors were split into a training set and a test set according to the result of a Kohonen's self-organizing map (SOM) or randomly. Multiple Linear Regression (MLR) models, support vector machine (SVM) models and two consensus models were built based on the training sets by 20 selected CORINA Symphony descriptors. RESULTS: All the models showed a good prediction of pIC50. The correlation coefficients of all the models were more than 0.7 on the test set. For the training set of consensus Model C1, which performed better than other models, the correlation coefficient(r) achieved 0.909 on the training set, and 0.804 on the test set. CONCLUSION: The selected molecular descriptors show that hydrogen bond acceptor, atom charges and electronegativities (especially π atom) were important in predicting the activity of HIV-1 integrase LEDGF/p75-IN inhibitors.

Geospatial, Clinical, and Social Determinants of Hospital Readmissions.

Unnecessary hospital readmissions increase patient burden, decrease health care quality and efficiency, and raise overall costs. This retrospective cohort study sought to identify high-risk patients who may serve as targets for interventions aiming at reducing hospital readmissions. The authors compared geospatial, social demographic, and clinical characteristics of patients with or without a 90-day readmission. Electronic health records of 42 330 adult patients admitted to 2 Midwestern hospitals during 2013 to 2016 were used, and logistic regression was performed to determine risk factors for readmission. The 90-day readmission percentage was 14.9%. Two main groups of patients with significantly higher odds of a 90-day readmission included those with severe conditions, particularly those with a short length of stay at incident admission, and patients with Medicare but younger than age 65. These findings expand knowledge of potential risk factors related to readmissions. Future interventions to reduce hospital readmissions may focus on the aforementioned high-risk patient groups.

Classification of Cyclooxygenase-2 Inhibitors Using Support Vector Machine and Random Forest Methods.

This work reports the classification study conducted on the biggest COX-2 inhibitor data set so far. Using 2925 diverse COX-2 inhibitors collected from 168 pieces of literature, we applied machine learning methods, support vector machine (SVM) and random forest (RF), to develop 12 classification models. The best SVM and RF models resulted in MCC values of 0.73 and 0.72, respectively. The 2925 COX-2 inhibitors were reduced to a data set of 1630 molecules by removing intermediately active inhibitors, and 12 new classification models were constructed, yielding MCC values above 0.72. The best MCC value of the external test set was predicted to be 0.68 by the RF model using ECFP_4 fingerprints. Moreover, the 2925 COX-2 inhibitors were clustered into eight subsets, and the structural features of each subset were investigated. We identified substructures important for activity including halogen, carboxyl, sulfonamide, and methanesulfonyl groups, as well as the aromatic nitrogen atoms. The models developed in this study could serve as useful tools for compound screening prior to lab tests.

SAR study on inhibitors of GIIA secreted phospholipase A2 using machine learning methods.

GIIA secreted phospholipase A2 (GIIA sPLA2 ) is a potent target for drug discovery. To distinguish the activity level of the inhibitors of GIIA sPLA2 , we built 24 classification models by three machine learning algorithms including support vector machine (SVM), decision tree (DT), and random forest (RF) based on 452 compounds. The molecules were represented by CORINA descriptors, MACCS fingerprints, and ECFP4 fingerprints, respectively. The dataset was split into a training set containing 312 compounds and a test set containing 140 compounds by Kohonen's self-organizing map (SOM) strategy and a random strategy. A recursive feature elimination (RFE) method and an information gain (IG) method were used in the selection of molecular descriptors. Three favorable performing models were obtained. They were built by SVM algorithm with CORINA descriptors (Models 1A and 2A) and ECFP4 fingerprints (Model 10A). In the prediction of test set of Model 10A, the accuracy reached 90.71%, and the Matthews correlation coefficient (MCC) values reached 0.82. In addition, the 452 inhibitors were clustered into eight subsets by K-Means algorithm for analyzing their structural features. It was found that highly active inhibitors mainly contained indole scaffold or indolizine scaffold and four side chains.

Classification of HIV-1 Protease Inhibitors by Machine Learning Methods.

HIV-1 protease plays an important role in the processing of virus infection. Protease is an effective therapeutic target for the treatment of HIV-1. Our data set is based on a selection of 4855 HIV-1 protease inhibitors (PIs) from ChEMBL. A series of 15 classification models for predicting the active inhibitors were built by machine learning methods, including k-nearest neighors (K-NN), decision tree (DT), random forest (RF), support vector machine (SVM), and deep neural network (DNN). The molecular structures were characterized by (1) fingerprint descriptors including MACCS fingerprints and PubChem fingerprints and (2) physicochemical descriptors calculated by CORINA Symphony. The prediction accuracies of all of the models are more than 70% on the test set; the best accuracy of 83.07% was obtained by model 4A, which was built by the SVM method based on MACCS fingerprint descriptors. Nine consensus models were built with three kinds of different descriptors, which combined all of the machine learning methods using the "consensus prediction". Model C3a developed with MACCS fingerprint descriptors showed the highest accuracy on both training set (91.96%) and test set (83.15%). An external validation set including 35 989 compounds from DUD database and 239 active inhibitors from the recent literature was used to verify the performance of our model. The best prediction accuracy of 98.37% was obtained by model 3C, which was built by RF based on CORINA Symphony descriptors. In addition, from the analysis of molecular descriptors, it shows that the aromatic system and atoms related to hydrogen bonding provide important contributions to the bioactivity of PIs.

QSAR studies of the bioactivity of hepatitis C virus (HCV) NS3/4A protease inhibitors by multiple linear regression (MLR) and support vector machine (SVM).

In this study, quantitative structure-activity relationship (QSAR) models using various descriptor sets and training/test set selection methods were explored to predict the bioactivity of hepatitis C virus (HCV) NS3/4A protease inhibitors by using a multiple linear regression (MLR) and a support vector machine (SVM) method. 512 HCV NS3/4A protease inhibitors and their IC50 values which were determined by the same FRET assay were collected from the reported literature to build a dataset. All the inhibitors were represented with selected nine global and 12 2D property-weighted autocorrelation descriptors calculated from the program CORINA Symphony. The dataset was divided into a training set and a test set by a random and a Kohonen's self-organizing map (SOM) method. The correlation coefficients (r2) of training sets and test sets were 0.75 and 0.72 for the best MLR model, 0.87 and 0.85 for the best SVM model, respectively. In addition, a series of sub-dataset models were also developed. The performances of all the best sub-dataset models were better than those of the whole dataset models. We believe that the combination of the best sub- and whole dataset SVM models can be used as reliable lead designing tools for new NS3/4A protease inhibitors scaffolds in a drug discovery pipeline.