Advanced computational approaches for single-cell multi-omics integration

University Of Pittsburgh At Pittsburgh, Pittsburgh PA

Investigators

Abstract

Enter the text here that is the new abstract information for your application. Single-cell technology has significantly advanced precision medicine by comprehensively characterizing individual cells. Integrating single-cell data across multi-omics platforms and diverse cohorts holds the potential for discovering more systematic and robust biomarkers. However, while existing tools have addressed batch effects introduced by different studies, they often risk losing valuable information due to under- or over-correction of the data. To overcome these challenges, we propose innovative integrative analyses to efficiently combine data from multiple modalities and cohorts while minimizing data distortion caused by excessive batch effect correction. Over the next five years, we will focus on three main themes. Theme 1 will develop novel integration approaches for single-cell RNA-seq, ATAC-seq, and cytometry data, leveraging gene expression, chromatin accessibility, and protein profiling data. We will first perform single-cell clustering and marker detection within individual studies, followed by innovative integration using meta-p-value or meta-effect-size approaches. Theme 2 will design advanced algorithms specifically for spatial transcriptomics and proteomics data. By integrating multiple single-cell and spatial-omics datasets, the proposed algorithms will characterize the tissue microenvironment by spatial trajectory inference and cell-cell interaction analysis. Theme 3 will elucidate transcript variants in both bulk and single-cell long-read RNA-seq data. We will first design model-based algorithms tailored to the state-of-the-art long-read data for isoform quantification and structural variation analysis. These approaches will then be extended to multi-cohort integration to discover robust biomarkers at single-cell and isoform-specific resolution. Building on the PI's existing strong collaborations with wet-lab biologists, algorithms developed across these themes will be optimized and validated in broad biomedical applications, including studies of chronic liver diseases, hepatocellular carcinoma zonation, acute kidney injury, and early-life immune development. The novel insights gained into cellular heterogeneity, tissue microenvironments, and transcript variants promise to have significant translational impacts through the proposed software. The overarching mission of the PI’s lab is to design innovative computational methodologies tailored to cutting-edge high-throughput multi-omics data. These approaches will be applied to a wide range of biomedical systems, where feedback will inspire further algorithm development, ultimately contributing to the clinical implementation of precision medicine. The PI's lab is dedicated to advanced methodology development and biological applications, focusing on both bulk and single-cell data for biomarker discovery, isoform annotation, machine learning analysis, and multi-omics integration. These past successes, along with the collaborative environment, make the PI exceptionally well-suited to lead the proposed projects. Ultimately, this R35 award will strengthen the PI’s long-term commitment to mentoring biomedical trainees and developing generalizable computational tools for translational and clinical applications of human disease relevance.