Using an aligner other than the default one (minimap2)

[alexeigurevich]

The idea is inspired by the following twit

By default, QUAST utilizes the minimap2 aligner (before v.5 it was Nucmer). It is possible to adjust some alignment parameters via QUAST options (e.g., --min-identity) or by direct changes in the QUAST source code (e.g., here). However, switching to a completely different aligner is extremely challenging and requires both advanced Python skills and familiarity with the QUAST source code.

I envision three possible workarounds to make the aligner switch easier for the users. Let's discuss and select the best approach.

1. We can natively support an additional aligner, e.g., unimap if there is a high demand specifically for it. However, this doesn't seem a scalable strategy.
2. We can accept user-generated alignments (e.g., in the PAF or SAM format) in addition to raw assemblies and a reference genome. The problem here is that QUAST do some preprocessing of assemblies and the reference before the alignment, e.g., filters out short contigs (less than --min-contig, 500 bp by default). Still, we can try to postprocess the alignments accordingly.
3. We can accept a user-defined alignment command, e.g. /path/to/unimap --option1 --option2 <REF> <QUERY>, automatically substitute <REF> and <QUERY> placeholders inside the QUAST pipeline and execute the command. Of course, we would expect the alignment results in some standard format as in (2).

[1dayac]

I thought unimap is obsolete starting from minimap2.21 (or even 2.19) Basically all improvements of unimap are in minimap now чт, 20 янв. 2022 г., 4:24 Alexey Gurevich ***@***.***>:

…

The idea is inspired by the following twit <https://twitter.com/waywardsyintist/status/1482004451833569280> [image: image] <https://user-images.githubusercontent.com/620893/149977659-a26907fb-94e2-4fe5-8c79-19d76b9b776e.png> By default, QUAST utilizes the minimap2 aligner (before v.5 it was Nucmer). It is possible to adjust some alignment parameters via QUAST options (e.g., --min-identity) or by direct changes in the QUAST source code (e.g., here <https://github.com/ablab/quast/blob/release_5.1.0/quast_libs/ca_utils/align_contigs.py#L78>). However, switching to a completely different aligner is extremely challenging and requires both advanced Python skills and familiarity with the QUAST source code. I envision three possible workarounds to make the aligner switch easier for the users. Let's discuss and select the best approach. 1. We can natively support an additional aligner, e.g., unimap if there is a high demand specifically for it. However, this doesn't seem a scalable strategy. 2. We can accept user-generated alignments (e.g., in the PAF or SAM format) in addition to raw assemblies and a reference genome. The problem here is that QUAST do some preprocessing of assemblies and the reference before the alignment, e.g., filters out short contigs (less than --min-contig, 500 bp by default). Still, we can try to postprocess the alignments accordingly. 3. We can accept a user-defined alignment command, e.g. /path/to/unimap --option1 --option2 <REF> <QUERY>, automatically substitute <REF> and <QUERY> placeholders inside the QUAST pipeline and execute the command. Of course, we would expect the alignment results in some standard format as in (2). — Reply to this email directly, view it on GitHub <#201>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABKP3LP3DVRUGTBVTVDCBVTUW7IFJANCNFSM5MMGNLQQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>