Chaudhuri, A. A. | Search bioRxiv

bioRxiv2024

Mapping single-cell developmental potential in health and disease with interpretable deep learning

Single-cell RNA sequencing (scRNA-seq) has transformed our understanding of cell fate in developmental systems. However, identifying the molecular hallmarks of potency - the capacity of a cell to differentiate into other cell types - has remained challenging. Here, we introduce CytoTRACE 2, an interpretable deep learning framework for characterizing potency and differentiation states on an absolute scale from scRNA-seq data. Across 31 human and mouse scRNA-seq datasets encompassing 28 tissue types, CytoTRACE 2 outperformed existing methods for recovering experimentally determined potency levels and differentiation states covering the entire range of cellular ontogeny. Moreover, it reconstructed the temporal hierarchy of mouse embryogenesis across 62 timepoints; identified pan-tissue expression programs that discriminate major potency levels; and facilitated discovery of cellular phenotypes in cancer linked to survival and immunotherapy resistance. Our results illuminate a fundamental feature of cell biology and provide a broadly applicable platform for delineating single-cell differentiation landscapes in health and disease.

genomics↗

bioRxiv2019

Stool-derived eukaryotic RNA biomarkers for detection of high-risk adenomas

Background and aimsColorectal cancer (CRC) is the second leading cause of cancer related deaths in the United States. Mortality is largely attributable to low patient compliance with screening and a subsequent high frequency of late-stage diagnoses. Noninvasive methods, such as stool- or blood-based diagnostics could improve patient compliance, however, existing techniques cannot adequately detect high-risk adenomas (HRAs) and early-stage CRC. MethodsHere we apply cancer profiling using amplicon sequencing of stool-derived eukaryotic RNA for 275 patients undergoing prospective CRC screening. A training set of 154 samples was used to build a random forest model that included 4 feature types (differentially expressed amplicons, total RNA expression, demographic information, and fecal immunochemical test results). An independent hold out test set of 121 patients was used to assess model performance. ResultsWhen applied to the 121-patient hold out test set, the model attained a receiver operating characteristic (ROC) area under the curve (AUC) of 0.94 for CRC and a ROC AUC of 0.87 for CRC and HRAs. In aggregate, the model achieved a 91 % sensitivity for CRC and a 73% sensitivity for HRAs at an 89% specificity for all other findings (medium-risk adenomas, low-risk adenomas, benign polyps, and no findings on a colonoscopy). ConclusionCollectively, these results indicate that in addition to early CRC detection, stool-derived biomarkers can accurately and noninvasively identify HRAs, which could be harnessed to prevent CRC development for asymptomatic, average-risk patients.

cancer biology↗