Biology subjects

McMichael, J. F.

Publications and source records attributed to McMichael, J. F..

9 recordsLinked to original sources

bioRxiv2026

dgiLIT: A Method for Prioritization and AI Curation of Drug-Gene Interactions

IMPORTANCEThe Drug-Gene Interaction Database (DGIdb) has a long history of driving hypothesis generation for biomedical research through the careful curation of drug-gene interaction data from primary and secondary sources with supporting literature. Recent advances in large-language model (LLM) and artificial intelligence (AI) technologies have enabled new paradigms for knowledge extraction and biocuration. The accelerating growth of biomedical literature presents a significant challenge for maintaining up-to-date interaction data. With more than 38 million citations indexed in PubMed alone, new strategies must evolve to identify and incorporate new interaction data into DGIdb. OBJECTIVEIdentify new cost-effective AI curation strategies for incorporating new drug-gene interactions into DGIdb. METHODSWe present a methodology that leverages deterministic natural language processing techniques, existing harmonization frameworks, and AI-assisted curation to systematically narrow the literature space and identify new drug-gene interactions from published studies for inclusion in DGIdb. RESULTSWe demonstrate the use of lemmatization to prioritize a set of 100 abstracts containing high amounts of interaction words for downstream AI curation. From our set of abstracts, we were then able to identify 137 drug-gene interactions via an AI curation task, with 121 (88.3%) of these interactions being completely novel to DGIdb. A human expert evaluator reviewed this interaction set and was able to validate 134 of 137 (97.8%) interactions as being valid based on the text provided. CONCLUSIONTaken together, our results highlight a promising, cost-effective method of ingesting new interactions into DGIdb.

bioinformatics↗

bioRxiv2025

CompBio and MIRaS : A Multi-omic Analysis Platform Built on a Memory-Based Intelligence Engine

As molecular and cellular technologies have advanced, the need to analyze and interpret the resulting, often vast and multi-modal, data into actionable intelligence has become a rate-limiting factor in scientific advancement. While traditional knowledgebase-pathway tools and LLM-based AI tools both provide a degree of support in data interpretation, both exhibit limitations. Presented here is the CompBio multi-omic analysis platform built upon a novel memory-based intelligence engine, MIRaS. The system is not dependent on human-curated pathway knowledgebases, can perform statistical significance analysis, does not suffer from hallucination, and produces traceable results. Furthermore, substantial effort has been placed in the computer-human knowledge transfer components of the system to enable efficient interaction, interpretation, and learning. The system has undergone years of extensive testing, validation, and evolution in omics-based research with dozens of associated publications verifying its effectiveness. A systematic assessment of CompBio and a detailed description of the MIRaS method are provided.

bioinformatics↗

bioRxiv2025

Evaluating Language Models for Biomedical Fact-Checking: A Benchmark Dataset for Cancer Variant Interpretation Verification

Accurate interpretation of genomic variants is critical for precision oncology but remains slow and dependent on specialized expertise. Public knowledgebases such as the Clinical Interpretation of Variants in Cancer (CIViC) help by curating literature-backed variant interpretations in a structured form, yet verification and review have become major bottlenecks. Large language models (LLMs) offer a potential mechanism for accelerating biomedical claim verification, but their rapid turnover, variable availability, and known risks of unsupported reasoning require standardized and reproducible evaluation before integration into curation workflows. To address this, we developed CIViC-Fact, an expert-curated, full-text benchmark and evaluation framework. Domain experts linked structured cancer-variant claims to sentence-level evidence from source publications, including evidence from full-text articles, tables, and non-abstract sections that are commonly omitted from existing biomedical question-answering and scientific fact-checking datasets. Claim-verification reference labels were derived from CIViC records, revision histories and controlled data augmentation. A major finding of CIViC-Fact is that abstracts are insufficient for realistic biomedical claim verification. In the evaluated development subset of text-verifiable entries with full-text access, fewer than 30% could be fully validated from the abstract alone, highlighting the importance of full-text evaluation for biomedical curation. Upon the application of our fact-checking pipeline to newly submitted CIViC entries, after excluding entries requiring supplementary material or images for validation, automated retrieval successfully identified appropriate evidence for most cases (93%), supporting low-incremental-effort evaluation of future systems. Fine-tuning improved agreement with CIViC-Fact reference labels on the static benchmark, but larger general-purpose models performed better on a heterogeneous post-cutoff cohort. These findings support CIViC-Fact primarily as a reproducible framework for comparing evolving retrieval and verification systems rather than as validation of a single deployment-ready model. These findings suggest that, in a rapidly changing model landscape, the durable contribution is not a single optimized model but a reproducible benchmark framework that enables continual testing, model substitution, and lightweight updating through small high-quality few-shot exemplar sets.

bioinformatics↗

bioRxiv2021

Evolution of the open-access CIViC knowledgebase is driven by the needsof the cancer variant interpretation community

CIViC (Clinical Interpretation of Variants in Cancer; civicdb.org) is a crowd-sourced, public domain knowledgebase composed of literature-derived evidence characterizing the clinical utility of cancer variants. As clinical sequencing becomes more prevalent in cancer management, the need for cancer variant interpretation has grown beyond the capability of any single institution. With nearly 300 contributors, CIViC contains peer-reviewed, published literature curated and expert-moderated into structured data units (Evidence Items) that can be accessed globally and in real time, reducing barriers to clinical variant knowledge sharing. We have extended CIViCs functionality to support emergent variant interpretation guidelines, increase interoperability with other variant resources, and promote widespread dissemination of structured curated data. To support the full breadth of variant interpretation from basic to translational, including integration of somatic and germline variant knowledge and inference of drug response, we have enabled curation of three new evidence types (predisposing, oncogenic and functional). The growing CIViC knowledgebase distributes clinically-relevant cancer variant data currently representing >2500 variants in >400 genes from >2800 publications.

genomics↗

bioRxiv2021

Spatial drivers and pre-cancer populations collaborate with the microenvironment in untreated and chemo-resistant pancreatic cancer

Pancreatic Ductal Adenocarcinoma (PDAC) is a lethal disease with limited treatment options and poor survival. We studied 73 samples from 21 patients (7 treatment-naive and 14 treated with neoadjuvant regimens), analyzing distinct spatial units and performing bulk proteogenomics, single cell sequencing, and cellular imaging. Spatial drivers, including mutant KRAS, SMAD4, and GNAQ, were associated with differential phosphosignaling and metabolic responses compared to wild type. Single cell subtyping discovered 12 of 21 tumors with mixed basal and classical features. Trefoil factor family members were upregulated in classical populations, while the basal populations showed enhanced expression of mesenchymal genes, including VIM and IGTB1. Acinar-ductal metaplasia (ADM) populations, present in 95% of patients, with 46% reduction of driver mutation fractions compared to tumor populations, exhibited suppressive and oncogenic features linked to morphologic states. We identified coordinated expression of TIGIT in exhausted and regulatory T cells and Nectin receptor expression in tumor cells. Higher expression of angiogenic and stress response genes in dendritic cells compared to tumor cells suggests they have a pro-tumorigenic role in remodeling the microenvironment. Treated samples contain a three-fold enrichment of inflammatory CAFs when compared to untreated samples, while other CAF subtypes remain similar. A subset of tumor and/or ADM-specific biomarkers showed differential expression between treatment groups, and several known drug targets displayed potential cross-cell type reactivities. This resolution that spatially defined single cell omics provides reveals the diversity of tumor and microenvironment populations in PDAC. Such understanding may lead to more optimal treatment regimens for patients with this devastating disease. HIGHLIGHTSO_LIAcinar-ductal metaplasia (ADM) cells represent a genetic and morphologic transition state between acinar and tumor cells. C_LIO_LIInflammatory cancer associated fibroblasts (iCAFs) are a major component of the PDAC TME and are significantly higher in treated samples C_LIO_LIReceptor-ligand analysis reveals tumor cell-TME interactions through NECTIN4-TIGIT C_LIO_LITumor and ADM cell proteogenomics differ between treated and untreated samples, with unique and shared potential drug targets C_LI

cancer biology↗

bioRxiv2020

Accurate neoantigen prediction depends on mutation position relative to patient allele-specific MHC anchor location

Neoantigens are novel peptide sequences resulting from sources such as somatic mutations in tumors. Upon loading onto major histocompatibility complex (MHC) molecules, they can trigger recognition by T cells. Accurate neoantigen identification is thus critical for both designing cancer vaccines and predicting response to immunotherapies. Neoantigen identification and prioritization relies on correctly predicting whether the presenting peptide sequence can successfully induce an immune response. As the majority of somatic mutations are single nucleotide variants, changes between wildtype and mutated peptides are typically subtle and require cautious interpretation. A potentially underappreciated variable in neoantigen-prediction pipelines is the mutation position within the peptide relative to its anchor positions for the patients specific MHC molecules. While a subset of peptide positions are presented to the T-cell receptor for recognition, others are responsible for anchoring to the MHC, making these positional considerations critical for predicting T-cell responses. We computationally predicted high probability anchor positions for different peptide lengths for 328 common HLA alleles and identified unique anchoring patterns among them. Analysis of 923 tumor samples shows that 6-38% of neoantigen candidates are potentially misclassified and can be rescued using allelespecific knowledge of anchor positions. A subset of anchor results were orthogonally validated using protein crystallography structures. Representative anchor trends were experimentally validated using peptide-MHC stability assays and competition binding assays. By incorporating our anchor prediction results into neoantigen prediction pipelines, we hope to formalize, streamline and improve the identification process for relevant clinical studies. One Sentence SummaryNeoantigen prediction accuracy is significantly influenced by the mutation position within the neoantigen and its relative position to the patients allele-specific MHC anchor locations.

bioinformatics↗

bioRxiv2020

Integration of the Drug-Gene Interaction Database (DGIdb) with open crowdsource efforts

The Drug-Gene Interaction Database (DGIdb, www.dgidb.org) is a web resource that provides information on drug-gene interactions and druggable genes from various sources including publications, databases, and other web-based sources in one resource. These drug, gene, and interaction claims are normalized and grouped to identify aliases, merge concepts, and reduce redundancy. The information contained in this resource is available to users through a straightforward search interface, an application programming interface (API), and TSV data downloads. DGIdb 4.0 is the latest major update of this database. Seven new sources have been added, bringing the total number of sources included to 41. Of the previously aggregated sources, 15 have been updated. DGIdb 4.0 also includes improvements to the process of drug normalization and grouping of imported sources. Other notable updates include further development of automatic jobs for routine data updates, more sophisticated query scores for interaction search results, extensive manual curation of interaction source link outs, and the inclusion of interaction directionality. A major focus of this update was integration with crowd-sourced efforts, including leveraging the curation activities of Drug Target Commons, using Wikidata to facilitate term normalization, and integrating into NDEx for producing network representations.

bioinformatics↗

bioRxiv2019

CIViCpy: a Python software development and analysis toolkit for the CIViC knowledgebase

PurposePrecision oncology is dependent upon the matching of tumor variants to relevant knowledge describing the clinical significance of those variants. We recently developed the Clinical Interpretations for Variants in Cancer (CIViC; civicdb.org) crowd-sourced, expert-moderated, open-access knowledgebase, providing a structured framework for evaluating genomic variants of various types (e.g., fusions, SNVs) for their therapeutic, prognostic, predisposing, diagnostic, or functional utility. CIViC has a documented API for accessing CIViC records: Assertions, Evidence, Variants, and Genes. Third-party tools that analyze or access the contents of this knowledgebase programmatically must leverage this API, often reimplementing redundant functionality in the pursuit of common analysis tasks that are beyond the scope of the CIViC web application.\n\nMethodsTo address this limitation, we developed CIViCpy (civicpy.org), a software development kit (SDK) for extracting and analyzing the contents of the CIViC knowledgebase. CIViCpy enables users to query CIViC content as dynamic objects in Python. We assess the viability of CIViCpy as a tool for advancing individualized patient care by using it to systematically match CIViC evidence to observed variants in patient cancer samples.\n\nResultsWe used CIViCpy to evaluate variants from 59,437 sequenced tumors of the AACR Project GENIE dataset. We demonstrate that CIViCpy enables annotation of >1,200 variants per second, resulting in precise variant matches to CIViC level A (professional guideline) or B (clinical trial) evidence for 38.6% of tumors.\n\nConclusionsThe clinical interpretation of genomic variants in cancers requires high-throughput tools for interoperability and analysis of variant interpretation knowledge. These needs are met by CIViCpy, an SDK for downstream applications and rapid analysis. CIViCpy (civicpy.org) is fully documented, open-source, and freely available online.

bioinformatics↗

bioRxiv2019

The CIViC knowledge model and standard operating procedures for curation and clinical interpretation of variants in cancer

Manually curated variant knowledgebases and their associated knowledge models are serving an increasingly important role in distributing and interpreting variants in cancer. These knowledgebases vary in their level of public accessibility, and the complexity of the models used to capture clinical knowledge. CIViC (Clinical Interpretations of Variants in Cancer - www.civicdb.org) is a fully open, free-to-use cancer variant interpretation knowledgebase that incorporates highly detailed curation of evidence obtained from peer-reviewed publications. Currently, the CIViC knowledge model consists of four main components: Genes, Variants, Evidence Items, and Assertions. Each component has an associated knowledge model and methods for curation. Gene and Variant data contextualize the genomic region(s) involved in the clinical statement. Evidence Items provide structured associations between variants and their clinically predictive/therapeutic, prognostic, diagnostic, predisposing, and functional implications. Finally, CIViC Assertions summarize collections of CIViC Evidence Items for a specific Disease, Variant, and Clinical Significance with incorporation of clinical and technical guidelines. Here we present the CIViC knowledge model, curation standard operating procedures, and detailed examples to support community-driven curation of cancer variants.

bioinformatics↗