BAMs created by hapdiff are slow to load in IGV

[riyasj327]

Hi,

Just wanted to check if you know why the BAMs produced by Hapdiff by aligning the assembly to the reference genome using minimap2 is taking too much time to load in IGV and also to zoom into a particular region.

I am using hapdup_dual fastas as input to call variants from and the reference genome is hg38/T2T and then loading the hapdiff_mat/pat.bam produced by hapdiff to view in IGV.

It takes too much time to load and is significantly slower. I am using IGV 2.19.2.

Any thoughts on this would be really helpful!

thanks,
Riya

[mikolmogorov]

Hi Riya,

Slower compared to what? IGV could be fairly slow if you are looking at large regions (e.g. 100kb+). Bam format is standardized, so there should be nothing unusual about hapdiff output. Do other genome assembly alignments not produced by hapdiff visualize faster?

Misha

[riyasj327]

sorry for the confusion! I meant it was slower compared to the other BAMs not produced by hapdiff, yes!

So I did try manually running a minimap using the assemblies with the default flags and those files are comparatively easier to use with IGV. So I suspect is it some of the flags and values used by minimap in hapdiff causing the issue?

In hapdiff I see the minimap2 command for BAM generation is CL:minimap2 -ax asm20 -B 2 -E 3,1 -O 6,100 --cs -t 36 -K 5G whereas when I used the default assembly alignment command https://github.com/lh3/minimap2?tab=readme-ov-file#full-genomeassembly-alignment minimap2 -ax asm5 ref.fa asm.fa > aln.sam # assembly to assembly/ref alignment the bams are much easier to use (not taking too long to load and navigate). @mikolmogorov

[mikolmogorov]

Thanks - it is likely due to hapdiff alignments being longer, compared to minimap2 with default parameters. We optimized the parameters to span longer indels or regions with increased reference heterozygosity.

[riyasj327]

Okay, thank you so much for your response! @mikolmogorov