-fast option triggers .bca format required

[pentamorfico]

Description

Running reseek -search with the -fast option against the uniprot50 database fails with a fatal error stating that the database must be in .bca, even though the db is in .bca format. The same command works correctly when using -sensitive or -verysensitive.

This behavior is not documented and is unexpected, as the database works with other sensitivity modes.

Reseek has been built from source from the latest release

---

Command used

reseek -search < input > /mnt/diskdb/reseek/uniprot50 -fast

---

Observed behavior

---Fatal error---
.bca format required for -db

[rcedgar]

This command line is incorrect. Reseek does not support redirection for input or output. The database should be specified by the -db option. Reseek requires a known "STRUCTS" extension to determine the type of file .bca .pdb .cif .files etc.

[pentamorfico]

Hi Robert,

Apologies for the typo before, the command I used was:

reseek -search <input.pdb> -db /mnt/diskdb/reseek/uniprot50 -output <output.tsv> -fast

With the folder /mnt/diskdb/reseek/uniprot50 containing the .bca files downloaded from https://doi.org/10.5281/zenodo.14063905

[rcedgar]

ah ok. internally reseek swaps query and db in fast mode, please try converting the query to bca and see if that fixes it.

[rcedgar]

forresten hej København :-)

[pentamorfico]

Hi Robert, I still get the same issue:

reseek -search test.bca -db /mnt/diskdb/reseek/uniprot50 -output test.tsv -fast

reseek v2.8.i86linux64 []
(C) Copyright 2024 Robert C. Edgar

[search test.bca]


search.cpp(80): 

reseek -search test.bca -db /mnt/diskdb/reseek/uniprot50 -output test.tsv -fast

---Fatal error---
.bca format required for -db

ls /mnt/diskdb/reseek/uniprot50

resdb10.bca  resdb12.bca  resdb14.bca  resdb16.bca  resdb18.bca  resdb1.bca   resdb21.bca  resdb23.bca  resdb25.bca  resdb27.bca  resdb29.bca  resdb30.bca  resdb32.bca  resdb34.bca  resdb36.bca  resdb38.bca  resdb3.bca   resdb41.bca  resdb43.bca  resdb45.bca  resdb47.bca  resdb49.bca  resdb50.bca  resdb52.bca  resdb54.bca  resdb6.bca  resdb8.bca
resdb11.bca  resdb13.bca  resdb15.bca  resdb17.bca  resdb19.bca  resdb20.bca  resdb22.bca  resdb24.bca  resdb26.bca  resdb28.bca  resdb2.bca   resdb31.bca  resdb33.bca  resdb35.bca  resdb37.bca  resdb39.bca  resdb40.bca  resdb42.bca  resdb44.bca  resdb46.bca  resdb48.bca  resdb4.bca   resdb51.bca  resdb53.bca  resdb5.bca   resdb7.bca  resdb9.bca

forresten hej København :-)

I’m currently in Spain, escaping the Copenhagen winter! Hi from Spain! Hope everything is going good 🙌

[rcedgar]

Ola Espana. I failed to escape several of those Copenhagen winters. One day I woke up in my apartment in Svanemoellen, the central heating was broken and the water in my glass next to the bed was frozen.

/mnt/diskdb/reseek/uniprot50 is missing .bca extension

[rcedgar]

Oh sorry, I overlooked uniprot50 is a directory name. Possibly you need a trailing slash so that it is recognized as such. If not, next theory is that a directory is not supported, you need a single intact bca as the db. You can make one using reseek -convert uniprot50/ -bca -uniprot50.bca

[rcedgar]

That last theory is correct. Reseek needs random access to db structures in the second pass after filtering. A better error message would be In fast mode the db must be a single .bca file, I'll flag this for v3.

[rcedgar]

@pentamorfico I'm glad to see users trying reseek at AFDB scale, I'd love to learn more, email me if you can share more about your project. Also I'm working on "AFDB100" in bca format, i.e. all AFDB without the 50% clustering, it will be roughly 4x slower. Let me know if that would be helpful and I can share the data.