hh-suite icon indicating copy to clipboard operation
hh-suite copied to clipboard

Databases link not found

Open jchen634 opened this issue 11 months ago • 4 comments

Hi, I am trying to find the databases for running HHpred locally, however the databases link http://ftp.tuebingen.mpg.de/pub/protevo/toolkit/databases/hhsuite_dbs posted in Readme is not found. Pfam is available, where can I get access to use PDB, COG_KOG, NCBI, and SMART. Is it possible to do all the 5 databases at the same time? Thank you so much!

jchen634 avatar Mar 12 '25 21:03 jchen634

Seems like the new URL is http://ftp.tuebingen.mpg.de/pub/ebio/protevo/toolkit/databases/hhsuite_dbs/

milot-mirdita avatar Mar 13 '25 14:03 milot-mirdita

Thank you so much! Do you happen to have the SMART database and can I use 5 databases at the same time when running locally?

jchen634 avatar Mar 13 '25 19:03 jchen634

I would recommend to do separate searches. I think there are some unresolved issues with searching multiple databases at the same time. This is (i think) the cause of the occasional AlphaFold homology search crash, but we don't have any grant funding for hh-suite support, so we never investigated this.

milot-mirdita avatar Mar 14 '25 05:03 milot-mirdita

Thank you! I am using PDB database only and comparing the results from the HHpred website vs running it locally. This is my code for local hhpred: uni30_dir="/scratch/hhpred_dbs/UniRef30_2023_02/UniRef30_2023_02" pdb70_dir="/scratch/hhpred_dbs/pdb70_from_mmcif_2025-01-03/pdb70"

hhblits -cpu 8 -i "$file" -n 3 -cov 20 -e 1e-3 -qid 0 -maxmem 30 -o "$scratch_dir/temp/$r_name.hhr" -oa3m "$scratch_dir/temp/$r_name.a3m" -d "$uni30_dir"

hhsearch -cpu 8 -maxmem 30 -i "$scratch_dir/temp/$r_name.a3m" -o "$scratch_dir/hhpred_local_contxt/$r_name.hhr" -oa3m "$scratch_dir/hhpred_local_contxt/$r_name.a3m" -blasttab "$scratch_dir/hhpred_local_contxt/$r_name.txt" -d "$pdb70_dir" -p 20 -Z 10000 -z 1 -b 1 -B 10000 -ssm 2 -sc 1 -seq 1 -dbstrlen 10000 -norealign -maxres 32000 -contxt /home/groups/software/data/context_data.crf

And the result is: It's weird that the SS are all 0.0 Image

The results from HHpred website seem to have better scores:

Image

Do you know why and is there any way to improve the local one? Thank you so much!

jchen634 avatar Apr 02 '25 00:04 jchen634