How does CPhasing use proximity ligation paths with PoreC and HiFi-C data

[ScottGeib-USDA]

Hi, this is a great tool. I was trying to understand how it uses the data. Is CPhasing able to utilize the multiple connections between long read proximity ligation (HiC ) data, or does it only benefit from the longer component segments in the data (the reads are longer so they contain more variant information)? Are all connections considered pairwise, or is it able to consider multiple connections stringing a single chromosome together (all component segments in a single poreC/HiFi-C read should be from the same chromosome molecule (usually)), which could help drive separation into phase blocks?

Thanks!

[wangyibin]

Hi,
Sorry for the delayed response, and thank you for your insightful question!

You raise an excellent point — Pore-C/HiFi-C indeed captures multi-way connections, which can help in separating contigs into their respective chromosomes. Ideally, the multi-way connections from a single read should originate from the same chromosome, providing strong linkage information for assembling contigs into chromosome-scale scaffolds.

However, in practice, due to limitations of current mapping algorithms and the inherent noise in 3C-based technologies, segments from a single read may sometimes align to different chromosomes, particularly in regions where homologous chromosomes are highly similar.

C-Phasing addresses this by leveraging both pairwise contacts and high-order concatemers. Specifically, we construct a hypergraph to represent high-order concatemers, while using pairwise contacts to identify and filter out h-trans errors (spurious connections between homologous chromosomes). After removing these errors, we apply a hypergraph-based community detection algorithm to cluster contigs accurately into chromosomes.

Best,
Yibin