seqan / lambda

LAMBDA – the Local Aligner for Massive Biological DatA
https://github.com/seqan/lambda
Other
77 stars 20 forks source link

LAMBDA_LONG_PROTEIN_SUBJ_SEQS DB build #115

Closed hildebra closed 6 years ago

hildebra commented 6 years ago

Hey, I downloaded the latest lambda (1.9.4) and rebuilt my indexes. However, now I get the problem that all these indexes are not useable:

LAMBDA - the Local Aligner for Massive Biological DatA

Version 1.9.4

Reading index properties...

ERROR: The following exception was thrown while reading the index: "Your lambda executable was built with LAMBDA_LONG_PROTEIN_SUBJ_SEQS, but the index was created by an executable that was built without it.You need to recreate the index or rebuild Lambda." Make sure the directory exists and is readable; recreate the index and try again. If the problem persists, report an issue at https://github.com/seqan/lambda/issues and include this output, as well as the output of lambda2 --version, thanks! system call /g/bork3/home/hildebra/dev/Perl/reAssemble2Spec//bin/lambda2-1.9.4-Linux-x86_64/bin/./lambda2 searchn -t 24 --percent-identity 75 -n 200 -e 1e-12 --seed-offset 7 --seed-length 14 --seed-delta 1 -b -3 --filter-putative-duplicates on -q /g/scb/bork/hildebra/Tamoc/GCglobal/GGC13001/ribos/reads_SSU.fna -i /g/bork3/home/hildebra/DB/MarkerG/PR2//gb203_pr2_all_10_28_99p.fasta.lambda -o /MF//GGC13001ITS//LCA/SSUriboRun_bl.0.m8.gz

The index was built using this command: /g/bork3/home/hildebra/dev/Perl/reAssemble2Spec//bin/lambda2-1.9.4-Linux-x86_64/bin/./lambda2 mkindexn -t 16 -d /g/bork3/home/hildebra/DB/MarkerG/PR2//gb203_pr2_all_10_28_99p.fasta

How can I fix this? I'm not even using a protein DB. many thanks, Falk

hildebra commented 6 years ago

/g/bork3/home/hildebra/dev/Perl/reAssemble2Spec//bin/lambda2-1.9.4-Linux-x86_64/bin/./lambda2 --version lambda2 version: 1.9.4 SeqAn version: 2.4.0

rrahn commented 6 years ago

Hi the maintainer of the application is currently on vacation, so it may take him a week to get back to your issue.

h-2 commented 6 years ago

@hildebra Sorry about that, the 1.9.x branch doesn't yet have integration tests (because results still change slightly between releases) so that issues wasn't detected.

I will fix it next week, but if you are impatient, you can just comment out this entire if-clause and rebuild: https://github.com/seqan/lambda/blob/master/src/search_algo.hpp#L169

hildebra commented 6 years ago

Hey, yes would be great to have it working again. Btw, what are the expected performance gains between the ss4 and normal binary, that are in 1.9.3?

h-2 commented 6 years ago

Sorry for the long silence, development will pick up again next week!

Btw, what are the expected performance gains between the ss4 and normal binary, that are in 1.9.3?

I have measured around 30% for short reads, ~5% for long reads, but I would be interested in hearing your experience in this reagard, too!

h-2 commented 6 years ago

should be fixed in current master