Reproduce “$\textit{BlockFormer}$ : Transformer-based inference from genomic contact maps” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Eloïse Touron, Pedro L. C. Rodrigues, Julyan Arbel, Nelle Varoquaux, Michael Arbel
BlockFormer localizes centromeres in Hi-C maps with varying block structures by learning from simulated maps and encoding each patch’s block identity and within-block position. It applies one trained model across chromosomes instead of fitting anew for each genome. On a 32-kb S. cerevisiae map, its first-pass normalized error was 0.26 versus Centurion’s 0.62; results vary across species and chromosome sizes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.