Search bioRxiv⌕ Search

Biology subjects

Swisher, E.

Publications and source records attributed to Swisher, E..

2 recordsLinked to original sources

Diverse Patterns of Allele-Specific Expression in Healthy Human Tissues

Differences in gene sequence and gene expression underlie variation in traits. However, even monozygotic twins do not express their genes in the same way, develop divergence in traits, and succumb to distinct chronic diseases. During development, epigenetic silencing programs cause diversity in allele expression, resulting in differences in traits and chronic disease risk. To quantify human autosomal allele expression between individuals, we analyzed human allele-specific expression data from the GTEx project. For hundreds of genes, some individuals will express the gene biallelically, while many others may only express one allele or extreme bias towards one allele. We found gene-specific patterns of interindividual variation in allele bias. We found that some individuals have more genome-wide monoallelic/biased expression than others. Individuals also had distinct combinations of allele expression bias. These differences can underlie variation in traits, idiopathic or incompletely penetrant traits/diseases, and chronic diseases. Significance/ImpactAllele-specific expression can affect cancer, immune response, and genetic disease. This work reveals 1) gene-specific multimodal patterns of interindividual variation in allele bias, 2) that individuals can maintain bias across tissues, and 3) that different individuals have distinct combinations of silenced alleles. These different patterns and the weighted classifications demonstrate how allele bias manifests between individuals; there are individuals and tissues with more biased/non-Mendelian expression and some tissues have more age-related changes in which alleles are silenced.

genomics↗

HRDPath: An Explainable Multi-Model Deep Learning Architecture for Predicting Homologous Recombination Deficiency from Histopathology Images

Homologous recombination deficiency (HRD) is a critical biomarker for guiding treatment decisions in high-grade serous tubo-ovarian carcinoma (HGSOC), a cancer with few reliable biomarkers. However, existing genomic-based tests for HRD are variable, expensive, and time-consuming. To this end, we developed HRDPath, a novel patient-level deep learning architecture that combines the strengths of two complementary models with a multi-task design, to predict genomically derived HRD status from whole slide images in HGSOC. HRDPath was comprehensively validated across three datasets and benchmarked against leading deep learning models. It achieved an AUC of 0.846, surpassing previously reported H&E-based HRD prediction results for HGSOC images by 0.09, and for the first time, reporting a specificity of 0.938, where accuracy significantly increased when multiple slides per patient were used. Our proposed patient-level approach and interpretability pipeline enhance model trustworthiness and reveal important clinical and biological insights into HRD-positive cancers, highlighting the associated morphological and pathological changes at the cellular and tissue levels. HRDPath is a potentially accessible and scalable digital biomarker that could improve ovarian cancer diagnosis and therapy selection.

bioinformatics↗