⚠️ new preprint: Chromatin accessibility of primary cancers informs regional mutagenesis in metastases through multi-scale deep learning. led by @hanlijiang13.bsky.social @moleculargenetics.bsky.social @oicr.on.ca www.biorxiv.org/content/10.6... [1/3]
www.biorxiv.org
New preprint from our lab out "Pairwise genomic alterations identify prognostic tumor states in multiple cancer types". Most genomic models of cancer prognosis focus on single alterations. Here, we asked what can be learned from combinations of events [1/4] doi.org/10.64898/202...
We introduce CAMM, a multi-scale interpretable deep learning framework that models regional SNV/indel density, helps map the most important chromatin features contributing, and points out mutation-enriched coding and non-coding loci. github.com/reimandlab/C... [3/3]
By integrating WGS data from primary and metastatic cancers with epigenomics profiles, we show that even in metastatic cancers, regional mutagenesis remains strongly informed by the chromatin accessibility of the primary tumor cell of origin. [2/3]
Using genomic and clinical data from nearly 10,000 primary tumors, we identified 57 prognostic genomic interactions (PGIs): pairs of alterations whose joint status was associated with patient outcome beyond either alteration alone. [2/4]
These PGIs refined prognostic stratification and were linked to distinct transcriptomic programs, while also highlighting candidate genes within recurrent copy-number regions and supported functional dependencies. [3/4]
This study builds on PACIFIC, our computational method for detecting clinically important multi-omics interactions (aacrjournals.org/mcr/article/...). Here, we extend this interaction-based framework to discovery of prognostic signatures defined by pairs of genomic alterations. [4/4]