segmentation fault at make_msa.sh

[yoavatsmonraz]

So I have been trying for a while now to get RoseTTAFold to work with no actual success besides a 20 amino acid peptide that I tried just for the sake of trouble shooting my current issue. The error I'm getting with both run scripts (e2e and pyrosetta) is this : "line 26: 869232 Segmentation fault (core dumped) $HHBLITS -i $prev_a3m -oa3m $tmp_dir/t000_.$e.a3m -e $e -v 2" I tried using both the conda static hhblits and a binary of hhblits I compiled from source as suggested in the github and got the same issue in both cases. I tried changing the available number of threads from 1-8 and that didn't affect as well as the max available number of ram from 4-500 GB (again with the same error recurring). Ran an md5sum on both the bfd and uniref databases only to verify that they're accurate and they are indeed correct. Important to note here - this issue only happens when I query the bfd database and not the uniref one, so apparently size of the db might be an issue here. Tried different values of -maxres including 15000, 2000,1500,500 and 50 which lead to another set of issues in the lower numbers and segfaulted as described for 2000 and 15000. I have a commercial lic. of pyrosetta and it works flawlessly in any other task so I don't think its a licensing issue either. I've also tried reversing the order of the values in line 26 of make_msa.sh and that didn't help either, so I truly have no idea why this is happening. Please advise,

[pguillem]

Its just insane that these tools (HHsuite) are coded to work with 1.5TB plaintext files. A relational database would solve A LOT of issues.

[275145]

So I have been trying for a while now to get RoseTTAFold to work with no actual success besides a 20 amino acid peptide that I tried just for the sake of trouble shooting my current issue. The error I'm getting with both run scripts (e2e and pyrosetta) is this : "line 26: 869232 Segmentation fault (core dumped) $HHBLITS -i $prev_a3m -oa3m $tmp_dir/t000_.$e.a3m -e $e -v 2" I tried using both the conda static hhblits and a binary of hhblits I compiled from source as suggested in the github and got the same issue in both cases. I tried changing the available number of threads from 1-8 and that didn't affect as well as the max available number of ram from 4-500 GB (again with the same error recurring). Ran an md5sum on both the bfd and uniref databases only to verify that they're accurate and they are indeed correct. Important to note here - this issue only happens when I query the bfd database and not the uniref one, so apparently size of the db might be an issue here. Tried different values of -maxres including 15000, 2000,1500,500 and 50 which lead to another set of issues in the lower numbers and segfaulted as described for 2000 and 15000. I have a commercial lic. of pyrosetta and it works flawlessly in any other task so I don't think its a licensing issue either. I've also tried reversing the order of the values in line 26 of make_msa.sh and that didn't help either, so I truly have no idea why this is happening. Please advise, @yoavatsmonraz Have you solved the problem? I think I'm having the same problem as you, thanks in advance for your advice

[Wes-V3]

I got the same problem. Wish there were any solution.