To view the seminar recording, please click here

Talk Title:

Statistical approaches for deciphering gene programs in cancer cells across studies and mutational signatures across ancestries

Abstract:

Single-cell assays provide valuable insights into the biological processes active in cancer cells within and across tumor samples. However, the majority of single cell RNA-seq (scRNA-seq) studies only profile a relatively small number of tumors and do not fully capture the heterogeneity present across the full spectrum of disease. We developed a novel meta-analysis approach which first uses our factorization tool called Celda to cluster genes into co-expression modules across cancer cells and then uses mixed effect linear models to assess the association of each module with clinical and technical variables. Cancer cells from 127 lung tumors were collected, annotated, and analyzed from 12 scRNA-seq datasets. Exploration of modules revealed a high degree of lineage plasticity in early-stage tumors, gene programs activated in late-stage tumors, which modules were driven by underlying somatic alterations. The Campbell lab is also building novel statistical methods and software for characterization of mutational signatures in cancer. We characterized mutational signatures in a targeted sequencing dataset obtained from cBioPortal containing over 2,000 tumors from the Chinese population. In lung adenocarcinoma, the polycyclic aromatic hydrocarbon (PAH) signature SBS4 was significantly higher in males compared to females but not associated with smoking status. The UV-associated signature SBS7 was significantly lower in cutaneous melanomas from the Chinese population compared a cohort of tumors from MSK. Overall, these results add to our understanding of the mutational processes that contribute to tumors from the Chinese population.