Make the reference configurable

[leoisl]

This is a request from Floriane and Pedro from our meeting on 18/05/2022.

H37Rv is not always the reference they might want describe the variants with respect to, so the reference has to be configurable (e.g. through the parameters). Right now it is always forced to be H37Rv.

[mbhall88]

As long as they're aware that using different references will mean they won't be able to do SNP distances between isolates with different references

[leoisl]

Yeah, this is error prone. I am wondering if we just leave for the users to manage it or if we try to control for this issue, like adding the sha256 of the reference fasta to the header of the consensus, and when running tbpore cluster we can check if all consensus were generated against the same ref

…

On Thu, 19 May 2022, 01:34 Michael Hall, ***@***.***> wrote: As long as they're aware that using different references will mean they won't be able to do SNP distances between isolates with different references — Reply to this email directly, view it on GitHub <#35 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACACJD45W675NAAGBAQ3ZL3VKWEA7ANCNFSM5WJAVZCQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[mbhall88]

@iqbal-lab I think we need to stress this issue. I don't think we want to be detecting when they're providing consensus sequences based on different references...

[iqbal-lab]

It's called tbpore, I don't see why we need to support different references. Will follow up

[leoisl]

It is still a TB reference, but a different one. Sometimes they want to compare with an African (if I remember correctly) reference strain, not with H37Rv

[mbhall88]

Yep. But the genome will be a different size and therefore SNP distances won't be applicable

[leoisl]

So, should we reconsider this or should we close as not planned?

[mbhall88]

We can make it configurable - it shouldn't be too difficult right? And just make sure we add a disclaimer in the help menu and docs that when using a different reference you can't compare SNP distances to other samples using a different reference.

[leoisl]

It is not hard to make it configurable. We could also easily control that tbpore cluster is called with consensus that were generated from a same ref - just add the ref sha-256 or similar hash to the first fasta header of the consensus. Of course this control can't be enforced if the consensus was not made by tbpore but then we could just have a --ignore-ref-check flag

[mbhall88]

Sounds like a good idea to me