Introduction to the BioChemical Library (BCL): An Application-Based Open-Source Toolkit for Integrated Cheminformatics and Machine Learning in Computer-Aided Drug Discovery.

The BioChemical Library (BCL) cheminformatics toolkit is an application-based academic open-source software package designed to integrate traditional small molecule cheminformatics tools with machine learning-based quantitative structure-activity/property relationship (QSAR/QSPR) modeling. In this pedagogical article we provide a detailed introduction to core BCL cheminformatics functionality, showing how traditional tasks (e.g., computing chemical properties, estimating druglikeness) can be readily combined with machine learning. In addition, we have included multiple examples covering areas of advanced use, such as reaction-based library design. We anticipate that this manuscript will be a valuable resource for researchers in computer-aided drug discovery looking to integrate modular cheminformatics and machine learning tools into their pipelines.

Identification of Novel Allosteric Modulators of Metabotropic Glutamate Receptor Subtype 5 Acting at Site Distinct from 2-Methyl-6-(phenylethynyl)-pyridine Binding.

As part of the G-protein coupled receptor (GPCR) family, metabotropic glutamate (mGlu) receptors play an important role as drug targets of cognitive diseases. Selective allosteric modulators of mGlu subtype 5 (mGlu5) have the potential to alleviate symptoms of numerous central nervous system disorders such as schizophrenia in a more targeted fashion. Multiple mGlu5 positive allosteric modulators (PAMs), such as 1-(3-fluorophenyl)-N-((3-fluorophenyl)-methylideneamino)-methanimine (DFB), 3-cyano-N-(1,3-diphenyl-1H-pyrazol-5-yl)-benzamide (CDPPB), and 4-nitro-N-(1,3-diphenyl-1H-pyrazol-5-yl)-benzamide (VU-29), exert their actions by binding to a defined allosteric site on mGlu5 located in the seven-transmembrane domain (7TM) and shared by mGlu5 negative allosteric modulator (NAM) 2-methyl-6-(phenylethynyl)-pyridine (MPEP). Actions of the PAM N-{4-chloro-2-[(1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)methyl]phenyl}-2-hydroxybenzamide (CPPHA) are mediated by a distinct allosteric site in the 7TM domain different from the MPEP binding site. Experimental evidence confirms these findings through mutagenesis experiments involving residues F585 (TM1) and A809 (TM7). In an effort to investigate mGlu5 PAM selectivity for this alternative allosteric site distinct from MPEP binding, we employed in silico quantitative structure-activity relationship (QSAR) modeling. Subsequent ligand-based virtual screening prioritized a set of 63 candidate compounds predicted from a library of over 4 million commercially available compounds to bind exclusively to this novel site. Experimental validation verified the biological activity for seven of 63 selected candidates. Further, medicinal chemistry optimizations based on these molecules revealed compound VU6003586 with an experimentally validated potency of 174 nM. Radioligand binding experiments showed only partial inhibition at very high concentrations, most likely indicative of binding at a non-MPEP site. Selective positive allosteric modulators for mGlu5 have the potential for tremendous impact concerning devastating neurological disorders such as schizophrenia and Huntington's disease. These identified and validated novel selective compounds can serve as starting points for more specifically tailored lead and probe molecules and thus help the development of potential therapeutic agents with reduced adverse effects.

Quality control and integration of genotypes from two calling pipelines for whole genome sequence data in the Alzheimer's disease sequencing project.

The Alzheimer's Disease Sequencing Project (ADSP) performed whole genome sequencing (WGS) of 584 subjects from 111 multiplex families at three sequencing centers. Genotype calling of single nucleotide variants (SNVs) and insertion-deletion variants (indels) was performed centrally using GATK-HaplotypeCaller and Atlas V2. The ADSP Quality Control (QC) Working Group applied QC protocols to project-level variant call format files (VCFs) from each pipeline, and developed and implemented a novel protocol, termed "consensus calling," to combine genotype calls from both pipelines into a single high-quality set. QC was applied to autosomal bi-allelic SNVs and indels, and included pipeline-recommended QC filters, variant-level QC, and sample-level QC. Low-quality variants or genotypes were excluded, and sample outliers were noted. Quality was assessed by examining Mendelian inconsistencies (MIs) among 67 parent-offspring pairs, and MIs were used to establish additional genotype-specific filters for GATK calls. After QC, 578 subjects remained. Pipeline-specific QC excluded ~12.0% of GATK and 14.5% of Atlas SNVs. Between pipelines, ~91% of SNV genotypes across all QCed variants were concordant; 4.23% and 4.56% of genotypes were exclusive to Atlas or GATK, respectively; the remaining ~0.01% of discordant genotypes were excluded. For indels, variant-level QC excluded ~36.8% of GATK and 35.3% of Atlas indels. Between pipelines, ~55.6% of indel genotypes were concordant; while 10.3% and 28.3% were exclusive to Atlas or GATK, respectively; and ~0.29% of discordant genotypes were. The final WGS consensus dataset contains 27,896,774 SNVs and 3,133,926 indels and is publicly available.

Functional annotation of genomic variants in studies of late-onset Alzheimer's disease.

Motivation: Annotation of genomic variants is an increasingly important and complex part of the analysis of sequence-based genomic analyses. Computational predictions of variant function are routinely incorporated into gene-based analyses of rare-variants, though to date most studies use limited information for assessing variant function that is often agnostic of the disease being studied. Results: In this work, we outline an annotation process motivated by the Alzheimer's Disease Sequencing Project, illustrate the impact of including tissue-specific transcript sets and sources of gene regulatory information and assess the potential impact of changing genomic builds on the annotation process. While these factors only impact a small proportion of total variant annotations (∼5%), they influence the potential analysis of a large fraction of genes (∼25%). Availability and implementation: Individual variant annotations are available via the NIAGADS GenomicsDB, at https://www.niagads.org/genomics/ tools-and-software/databases/genomics-database. Annotations are also available for bulk download at https://www.niagads.org/datasets. Annotation processing software is available at http://www.icompbio.net/resources/software-and-downloads/. Supplementary information: Supplementary data are available at Bioinformatics online.

Testosterone Pathway Genetic Polymorphisms in Relation to Primary Open-Angle Glaucoma: An Analysis in Two Large Datasets.

Purpose: Sex hormones may be associated with primary open-angle glaucoma (POAG), although the mechanisms are unclear. We previously observed that gene variants involved with estrogen metabolism were collectively associated with POAG in women but not men; here we assessed gene variants related to testosterone metabolism collectively and POAG risk. Methods: We used two datasets: one from the United States (3853 cases and 33,480 controls) and another from Australia (1155 cases and 1992 controls). Both datasets contained densely called genotypes imputed to the 1000 Genomes reference panel. We used pathway- and gene-based approaches with Pathway Analysis by Randomization Incorporating Structure (PARIS) software to assess the overall association between a panel of single nucleotide polymorphisms (SNPs) in testosterone metabolism genes and POAG. In sex-stratified analyses, we evaluated POAG overall and POAG subtypes defined by maximum IOP (high-tension [HTG] or normal tension glaucoma [NTG]). Results: In the US dataset, the SNP panel was not associated with POAG (permuted P = 0.77), although there was an association in the Australian sample (permuted P = 0.018). In both datasets, the SNP panel was associated with POAG in men (permuted P ≤ 0.033) and not women (permuted P ≥ 0.42), but in gene-based analyses, there was no consistency on the main genes responsible for these findings. In both datasets, the testosterone pathway association with HTG was significant (permuted P ≤ 0.011), but again, gene-based analyses showed no consistent driver gene associations. Conclusions: Collectively, testosterone metabolism pathway SNPs were consistently associated with the high-tension subtype of POAG in two datasets.

Genetic Variation in Genes Underlying Diverse Dementias May Explain a Small Proportion of Cases in the Alzheimer's Disease Sequencing Project.

BACKGROUND/AIMS: The Alzheimer's Disease Sequencing Project (ADSP) aims to identify novel genes influencing Alzheimer's disease (AD). Variants within genes known to cause dementias other than AD have previously been associated with AD risk. We describe evidence of co-segregation and associations between variants in dementia genes and clinically diagnosed AD within the ADSP. METHODS: We summarize the properties of known pathogenic variants within dementia genes, describe the co-segregation of variants annotated as "pathogenic" in ClinVar and new candidates observed in ADSP families, and test for associations between rare variants in dementia genes in the ADSP case-control study. The participants were clinically evaluated for AD, and they represent European, Caribbean Hispanic, and isolate Dutch populations. RESULTS/CONCLUSIONS: Pathogenic variants in dementia genes were predominantly rare and conserved coding changes. Pathogenic variants within ARSA, CSF1R, and GRN were observed, and candidate variants in GRN and CHMP2B were nominated in ADSP families. An independent case-control study provided evidence of an association between variants in TREM2, APOE, ARSA, CSF1R, PSEN1, and MAPT and risk of AD. Variants in genes which cause dementing disorders may influence the clinical diagnosis of AD in a small proportion of cases within the ADSP.

Introducing COCOS: codon consequence scanner for annotating reading frame changes induced by stop-lost and frame shift variants.

SUMMARY: Reading frame altering genomic variants can impact gene expression levels and the structure of protein products, thus potentially inducing disease phenotypes. Current annotation approaches report the impact of such variants in the context of altered DNA sequence only; attributes of the resulting transcript, reading frame and translated protein product are not reported. To remedy this shortcoming, we present a new genetic annotation approach termed Codon Consequence Scanner (COCOS). Implemented as an Ensembl variant effect predictor (VEP) plugin, COCOS captures amino acid sequence alterations stemming from variants that produce an altered reading frame, such as stop-lost variants and small insertions and deletions (InDels). To highlight its significance, COCOS was applied to data from the 1000 Genomes Project. Transcripts affected by stop-lost variants introduce a median of 15 amino acids, while InDels have a more extensive impact with a median of 66 amino acids being incorporated. Captured sequence alterations are written out in FASTA format and can be further analyzed for impact on the underlying protein structure. AVAILABILITY AND IMPLEMENTATION: COCOS is available to all users on github: https://github.com/butkiem/COCOS. CONTACT: mariusz.butkiewicz@case.edu.

High-Throughput Screening Assay Datasets from the PubChem Database.

Availability of high-throughput screening (HTS) data in the public domain offers great potential to foster development of ligand-based computer-aided drug discovery (LB-CADD) methods crucial for drug discovery efforts in academia and industry. LB-CADD method development depends on high-quality HTS assay data, i.e., datasets that contain both active and inactive compounds. These active compounds are hits from primary screens that have been tested in concentration-response experiments and where the target-specificity of the hits has been validated through suitable secondary screening experiments. Publicly available HTS repositories such as PubChem often provide such data in a convoluted way: compounds that are classified as inactive need to be extracted from the primary screening record. However, compounds classified as active in the primary screening record are not suitable as a set of active compounds for LB-CADD experiments due to high false-positive rate. A suitable set of actives can be derived by carefully analysing results in often up to five or more assays that are used to confirm and classify the activity of compounds. These assays, in part, build on each other. However, often not all hit compounds from the previous screen have been tested. Sometimes a compound can be classified as 'active', though its meaning is 'inactive' on the target of interest as it is 'active' on a different target protein. Here, a curation process of hierarchically related confirmatory screens is illustrated based on two specifically chosen protein use-cases. The subsequent re-upload procedure into PubChem is described for the findings of those two scenarios. Further, we provide nine publicly accessible high quality datasets for future LB-CADD method development that provide a common baseline for comparison of future methods to the scientific community. We also provide a protocol researchers can follow to upload additional datasets for benchmarking.

Shared Genetic Etiology of Autoimmune Diseases in Patients from a Biorepository Linked to De-identified Electronic Health Records.

Autoimmune diseases represent a significant medical burden affecting up to 5-8% of the U.S. POPULATION: While genetics is known to play a role, studies of common autoimmune diseases are complicated by phenotype heterogeneity, limited sample sizes, and a single disease approach. Here we performed a targeted genetic association study for cases of multiple sclerosis (MS), rheumatoid arthritis (RA), and Crohn's disease (CD) to assess which common genetic variants contribute individually and pleiotropically to disease risk. Joint modeling and pathway analysis combining the three phenotypes were performed to identify common underlying mechanisms of risk of autoimmune conditions. European American cases of MS, RA, and CD, (n = 119, 53, and 129, respectively) and 1924 controls were identified using de-identified electronic health records (EHRs) through a combination of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) billing codes, Current Procedural Terminology (CPT) codes, medication lists, and text matching. As expected, hallmark SNPs in MS, such as DQA1 rs9271366 (OR = 1.91; p = 0.008), replicated in the present study. Both MS and CD were associated with TIMMDC1 rs2293370 (OR = 0.27, p = 0.01; OR = 0.25, p = 0.02; respectively). Additionally, PDE2A rs3781913 was significantly associated with both CD and RA (OR = 0.46, p = 0.02; OR = 0.32, p = 0.02; respectively). Joint modeling and pathway analysis identified variants within the KEGG NOD-like receptor signaling pathway and Shigellosis pathway as being correlated with the combined autoimmune phenotype. Our study replicated previously-reported genetic associations for MS and CD in a population derived from de-identified EHRs. We found evidence to support a shared genetic etiology between CD/MS and CD/RA outside of the major histocompatibility complex region and identified KEGG pathways indicative of a bacterial pathogenesis risk for autoimmunity in a joint model. Future work to elucidate this shared etiology will be key in the development of risk models as envisioned in the era of precision medicine.

DRUG-DRUG INTERACTION PROFILES OF MEDICATION REGIMENS EXTRACTED FROM A DE-IDENTIFIED ELECTRONIC MEDICAL RECORDS SYSTEM.

With age, the number of prescribed medications increases and subsequently raises the risk for adverse drug-drug interactions. These adverse effects lower quality of life and increase health care costs. Quantifying the potential burden of adverse effects before prescribing medications can be a valuable contribution to health care. This study evaluated medication lists extracted from a subset of the Vanderbilt de-identified electronic medical record system. Reported drugs were cross-referenced with the Kyoto Encyclopedia of Genes and Genomes DRUG database to identify known drug-drug interactions. On average, a medication regimen contained 6.58 medications and 2.68 drug-drug interactions. Here, we quantify the burden of potential adverse events from drug-drug interactions through drug-drug interaction profiles and include a number of alternative medications as provided by the Anatomical Therapeutic Chemical Classification System.

Pathway analysis by randomization incorporating structure-PARIS: an update.

MOTIVATION: We present an update to the pathway enrichment analysis tool 'Pathway Analysis by Randomization Incorporating Structure (PARIS)' that determines aggregated association signals generated from genome-wide association study results. Pathway-based analyses highlight biological pathways associated with phenotypes. PARIS uses a unique permutation strategy to evaluate the genomic structure of interrogated pathways, through permutation testing of genomic features, thus eliminating many of the over-testing concerns arising with other pathway analysis approaches. RESULTS: We have updated PARIS to incorporate expanded pathway definitions through the incorporation of new expert knowledge from multiple database sources, through customized user provided pathways, and other improvements in user flexibility and functionality. AVAILABILITY AND IMPLEMENTATION: PARIS is freely available to all users at https://ritchielab.psu.edu/software/paris-download CONTACT: jnc43@case.edu SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In Silico Functional Annotation of Genomic Variation.

This unit describes the concepts and practical techniques for annotating genomic variants in the human genome to estimate their functional significance. With the rapid increase of available whole exome and whole genome sequencing information for human studies, annotation techniques have become progressively more important for highlighting and prioritizing nucleotide variants and their potential impact on genes and other genetic constructs. Here, we present an overview of different types of variant annotation approaches and elaborate on their foundations, assumptions, and the downstream consequences of their use. Computational approaches and tools to assign annotations and to identify variants are reviewed. Further, the general philosophy of assigning potential function to a genetic change within the biological context of a disease is discussed.

Rapidly progressive Alzheimer's disease features distinct structures of amyloid-ß.

Genetic and environmental factors that increase the risk of late-onset Alzheimer disease are now well recognized but the cause of variable progression rates and phenotypes of sporadic Alzheimer's disease is largely unknown. We aimed to investigate the relationship between diverse structural assemblies of amyloid-ß and rates of clinical decline in Alzheimer's disease. Using novel biophysical methods, we analysed levels, particle size, and conformational characteristics of amyloid-ß in the posterior cingulate cortex, hippocampus and cerebellum of 48 cases of Alzheimer's disease with distinctly different disease durations, and correlated the data with APOE gene polymorphism. In both hippocampus and posterior cingulate cortex we identified an extensive array of distinct amyloid-ß42 particles that differ in size, display of N-terminal and C-terminal domains, and conformational stability. In contrast, amyloid-ß40 present at low levels did not form a major particle with discernible size, and both N-terminal and C- terminal domains were largely exposed. Rapidly progressive Alzheimer's disease that is associated with a low frequency of APOE e4 allele demonstrates considerably expanded conformational heterogeneity of amyloid-ß42, with higher levels of distinctly structured amyloid-ß42 particles composed of 30-100 monomers, and fewer particles composed of < 30 monomers. The link between rapid clinical decline and levels of amyloid-ß42 with distinct structural characteristics suggests that different conformers may play an important role in the pathogenesis of distinct Alzheimer's disease phenotypes. These findings indicate that Alzheimer's disease exhibits a wide spectrum of amyloid-ß42 structural states and imply the existence of prion-like conformational strains.

Benchmarking ligand-based virtual High-Throughput Screening with the PubChem database.

With the rapidly increasing availability of High-Throughput Screening (HTS) data in the public domain, such as the PubChem database, methods for ligand-based computer-aided drug discovery (LB-CADD) have the potential to accelerate and reduce the cost of probe development and drug discovery efforts in academia. We assemble nine data sets from realistic HTS campaigns representing major families of drug target proteins for benchmarking LB-CADD methods. Each data set is public domain through PubChem and carefully collated through confirmation screens validating active compounds. These data sets provide the foundation for benchmarking a new cheminformatics framework BCL::ChemInfo, which is freely available for non-commercial use. Quantitative structure activity relationship (QSAR) models are built using Artificial Neural Networks (ANNs), Support Vector Machines (SVMs), Decision Trees (DTs), and Kohonen networks (KNs). Problem-specific descriptor optimization protocols are assessed including Sequential Feature Forward Selection (SFFS) and various information content measures. Measures of predictive power and confidence are evaluated through cross-validation, and a consensus prediction scheme is tested that combines orthogonal machine learning algorithms into a single predictor. Enrichments ranging from 15 to 101 for a TPR cutoff of 25% are observed.

BCL::EMAS--enantioselective molecular asymmetry descriptor for 3D-QSAR.

Stereochemistry is an important determinant of a molecule's biological activity. Stereoisomers can have different degrees of efficacy or even opposing effects when interacting with a target protein. Stereochemistry is a molecular property difficult to represent in 2D-QSAR as it is an inherently three-dimensional phenomenon. A major drawback of most proposed descriptors for 3D-QSAR that encode stereochemistry is that they require a heuristic for defining all stereocenters and rank-ordering its substituents. Here we propose a novel 3D-QSAR descriptor termed Enantioselective Molecular ASymmetry (EMAS) that is capable of distinguishing between enantiomers in the absence of such heuristics. The descriptor aims to measure the deviation from an overall symmetric shape of the molecule. A radial-distribution function (RDF) determines a signed volume of tetrahedrons of all triplets of atoms and the molecule center. The descriptor can be enriched with atom-centric properties such as partial charge. This descriptor showed good predictability when tested with a dataset of thirty-one steroids commonly used to benchmark stereochemistry descriptors (r² = 0.89, q² = 0.78). Additionally, EMAS improved enrichment of 4.38 versus 3.94 without EMAS in a simulated virtual high-throughput screening (vHTS) for inhibitors and substrates of cytochrome P450 (PUBCHEM AID891).

Iterative experimental and virtual high-throughput screening identifies metabotropic glutamate receptor subtype 4 positive allosteric modulators.

Activation of metabotropic glutamate receptor subtype 4 has been shown to be efficacious in rodent models of Parkinson's disease. Artificial neural networks were trained based on a recently reported high throughput screen which identified 434 positive allosteric modulators of metabotropic glutamate receptor subtype 4 out of a set of approximately 155,000 compounds. A jury system containing three artificial neural networks achieved a theoretical enrichment of 15.4 when selecting the top 2 % compounds of an independent test dataset. The model was used to screen an external commercial database of approximately 450,000 drug-like compounds. 1,100 predicted active small molecules were tested experimentally using two distinct assays of mGlu(4) activity. This experiment yielded 67 positive allosteric modulators of metabotropic glutamate receptor subtype 4 that confirmed in both experimental systems. Compared to the 0.3 % active compounds in the primary screen, this constituted an enrichment of 22 fold.

Identification of Metabotropic Glutamate Receptor Subtype 5 Potentiators Using Virtual High-Throughput Screening.

Selective potentiators of glutamate response at metabotropic glutamate receptor subtype 5 (mGluR5) have exciting potential for the development of novel treatment strategies for schizophrenia. A total of 1,382 compounds with positive allosteric modulation (PAM) of the mGluR5 glutamate response were identified through high-throughput screening (HTS) of a diverse library of 144,475 substances utilizing a functional assay measuring receptor-induced intracellular release of calcium. Primary hits were tested for concentration-dependent activity, and potency data (EC(50) values) were used for training artificial neural network (ANN) quantitative structure-activity relationship (QSAR) models that predict biological potency from the chemical structure. While all models were trained to predict EC(50), the quality of the models was assessed by using both continuous measures and binary classification. Numerical descriptors of chemical structure were used as input for the machine learning procedure and optimized in an iterative protocol. The ANN models achieved theoretical enrichment ratios of up to 38 for an independent data set not used in training the model. A database of approximately 450,000 commercially available drug-like compounds was targeted in a virtual screen. A set of 824 compounds was obtained for testing based on the highest predicted potency values. Biological testing found 28.2% (232/824) of these compounds with various activities at mGluR5 including 177 pure potentiators and 55 partial agonists. These results represent an enrichment factor of 23 for pure potentiation of the mGluR5 glutamate response and 30 for overall mGluR5 modulation activity when compared with those of the original mGluR5 experimental screening data (0.94% hit rate). The active compounds identified contained 72% close derivatives of previously identified PAMs as well as 28% nontrivial derivatives of known active compounds.