search

ComparativeGenomicsToolkit / hal

Hierarchical Alignment Format
Other
160 stars 38 forks source link

hal2chain for faster halLiftover #193

Open badoi opened 3 years ago

badoi commented 3 years ago

I saw that the hal2chain tool was shelved. Is it possible to make a chain based on the hal file to speed up cross-species alignment? This would provide a cross-species chain that would be faster than halLiftover which requires querying the hal file and bottlenecks simultaneous liftOvers. And if this chain works w/ the liftOver tool in place of the pairwise alignments, then that could work quite well with current tools that work w/ liftOver.

Current scenario:

halLiftover file.hal sourceSpeices input.bed targetSpecies output.bed

Example scenario:

hal2chain file.hal sourceSpeices targetSpecies > hal.sourceSpeicesToTargetSpecies.chain.gz
liftOver input.bed hal.sourceSpeicesToTargetSpecies.chain.gz output.bed unlifted.bed
diekhans commented 3 years ago

There is a document being written on how to build chains from HAL here:

https://github.com/ComparativeGenomicsToolkit/hal/blob/chaining-doc/doc/chaining-mapping.md

Please give feedback.

badoi commented 3 years ago

Thanks Mark! I'll test drive this chain building tool. Will get back to you on it.

DustinSokolowski commented 1 year ago

Hey @diekhans

This script works great to make a chain file compatible with transmap but it's actually backward for the normal convention.

So for normal liftOver or other tools (e.g., TOGA) the source species needs to be switched.

Best! Dustin

diekhans commented 1 year ago

This is a weird historic difference, but easy to correct by throwing in chainSwap.

Dustin Sokolowski @.***> writes:

Hey @diekhans

This script works great to make a chain file compatible with transmap but it's actually backward for the normal convention.

So for normal liftOver or other tools (e.g., TOGA) the source species needs to be switched.

Best! Dustin

-- Reply to this email directly or view it on GitHub: https://github.com/ComparativeGenomicsToolkit/hal/issues/193#issuecomment-1548805843 You are receiving this because you were mentioned.

Message ID: @.***>

DustinSokolowski commented 1 year ago

Hey!

It works great! I figured it would be good to put this here for anyone else getting weird results. and wondering what could be up.

Dustin