search

steineggerlab / foldseek

Foldseek enables fast and sensitive comparisons of large structure sets.
https://foldseek.com
GNU General Public License v3.0
821 stars 101 forks source link

Default value for -c in easy-cluster is wrong #299

Open chao9916 opened 4 months ago

chao9916 commented 4 months ago

Expected Behavior

In help page of easy-cluster module, default value for -c is 0.000. -c is for alignment (List matches above this fraction of aligned (covered) residues (see --cov-mode) (default: 0.0); higher coverage = more global alignment).

-c FLOAT List matches above this fraction of aligned (covered) residues (see --cov-mode) [0.000]

Current Behavior

But when I run easy-cluster module with default value for -c, the result is not the result with using 0.000 for -c. The value for -c should be about 0.800, which is same with default value for -c in cluster module.

Steps to Reproduce (for bugs)

foldseek easy-cluster /folder/to/inputdb /cluster/name tmp -e 1.000E+05 foldseek easy-cluster /folder/to/inputdb /cluster/name tmp -c 0.800 -e 1.000E+05

Foldssek Output (for bugs)

These two code give the same result. But the default value for -c is 0.000.

Context

The default value for -c in easy-cluster module shoud be changed to correct one in help page.

Your Environment

CentOS Linux 7 (Core) foldseek version: 8.ef4e960 from bioconda Also checked in newest version 9.427df8a from bioconda

agitter commented 1 day ago

I was also curious about this default value of 0.0. It is set here and does appear to be used from the small amount of the source I looked at. https://github.com/steineggerlab/foldseek/blob/d60563224ad2cf976af580cc48b1568bd35b4716/lib/mmseqs/src/commons/Parameters.cpp#L2312

I haven't run tests myself with the default value.

martin-steinegger commented 1 day ago

Easy-cluster uses a coverage of 0.8 while search doesn’t apply any coverage threshold in default. We need to update the help text for the easy-cluster.