Repository navigation
Usage of mmseqs dbs with metauk easy-predict and taxtocontig #97
Description
Activity
Hi, so sorry for not answering... I will try to have a look in a few days. Thank you for your patience.
Reacted by saracargnelutti01-opsHi again,
Thanks for the detailed report! Your suspicion sounds plausible.
We have just updated the bundled MMseqs2 to the latest version. Could you try downloading the latest precompiled binary from https://mmseqs.com/metaeuk/ and test it with your existing database? If this doesn't resolve the issue, could you please report back?
Thank you,
EliThank you so much! I'll try (maybe not that soon) and let you know.
Reacted by Eli Levy KarinHi! I'm back on metaeuk @elileka :)
I downloaded the binary:wget https://mmseqs.com/metaeuk/metaeuk-linux-avx2.tar.gz tar -xvzf metaeuk-linux-avx2.tar.gzI ran metaeuk easy-predict and taxtocontig with the db created using mmseqs from UniRef90 and it's now working!
Reacted by Eli Levy KarinNice! Thank you for reporting back :)
Reacted by saracargnelutti01-opsThank you for fixing the error @elileka
I'm keep on testing both mmseqs and metaeuk for the analysis of different datasets of putative eukaryotic bins. If I understood it correctly, MetaEuk is specifically tailored for eukaryotic sequences because it considers the eukaryotic exon-intron gene structure through creating spliced multi-exon proteins that are the actual queries aligned against the reference db; whereas MMseqs2 taxonomy evaluates extracted protein fragments individually. I'd like to ask what is the expected impact of this difference specifically in the context of taxonomic annotations.
I'm also curious to know what do you think about concatenating all contig sequences of a bin using a separator (https://github.com/metashot/awemags#taxonomic-classification) producing a fasta file with just one header in order to have just one taxonomic annotation for the entire bin.
I have created an mmeqs2 database using mmseqs2=18-8cc5c from the UniRef90 db and I'm trying to use it with metaeuk (version 7.bba0d80) easy-predict and taxtocontig modules. When I lounch these commands, metaeuk arises the following error:
No k-mer could be extracted for the database /db/UniRef90_mmseqs2/UniRef90
"Maybe the sequences length is less than 14 residues.
Error: Prefilter died
Error: First search died
Error: taxonomy died."
I suspect the problem is the recent (compared to metaeuk) version of mmseqs I used and so the resulting db is not compatible with metaeuk. Has someone went through the same issue? @elileka would you kindly have any suggestion?
I also have built a mmseqs db from the blast nr and I would like to use it with metaeuk as well.
Thanks in advance :)