How to reduce computational resource consumption

[tudouzi1111]

I hope this message finds you well. I am currently working on a genome analysis project inspired by your paper "Taurine pangenome uncovers a segmental duplication upstream of KIT associated with depigmentation in white-headed cattle." Specifically, I am following a similar approach by partitioning the entire pangenome into windows and calculating the Jaccard similarity within each window.

However, I have encountered a significant challenge during this process. The command odgi extract -i pig.renamed.sorted.og -r A#chr1:1-1000 -o chr1_1_1000.window_subgraph.og consumes a substantial amount of computational resources. Extracting just one subgraph from the full pangenome graph requires around 10 CPU cores. My pangenome consists of 30 genomes, each approximately 2G in size. As a result, extracting subgraphs in 1000 bp windows across the entire graph becomes computationally intensive and difficult to scale.

I was wondering if you might have any suggestions or alternative strategies to reduce the computational burden of this step, or to make the extraction process more efficient.

Thank you very much for your time and any insights you can share.

[AndreaGuarracino]

An alternative strategy, which I hope will become the main one, will be to use IMPG (https://github.com/pangenome/impg). In particular, impg query allows you to extract subsets of the pangenome. From these, you can create graphs and perform graph-based analysis.

IMPG will "soon" support direct graph generation, but for now, you can proceed by querying and PGGBing the impg query subsets.

[tudouzi1111]

An alternative strategy, which I hope will become the main one, will be to use IMPG (https://github.com/pangenome/impg). In particular, impg query allows you to extract subsets of the pangenome. From these, you can create graphs and perform graph-based analysis.

IMPG will "soon" support direct graph generation, but for now, you can proceed by querying and PGGBing the impg query subsets.

Thank you very much for your patient response. However, I still have some questions.
From what I’ve read, IMPG seems to be an efficient tool for processing multiple-sequence PAF alignment files.
However, my pangenome was built using VG, and I currently have a GAF file.
As far as I understand, GAF is not an alignment file — it doesn’t contain actual alignment information.
So I’m not sure how to apply IMPG to my GAF-based pangenome.