Segmentation fault when running createdb on alphafold mmcif_files directory about foldseek

[martin-steinegger]

Could you please share the input database with us?

[cjneely10]

Thank you for your response,

The input database is the set of all PDB mmcif files downloaded using rsync and then unzipping. I tried performing a fresh download/unzip:

rsync --progress -rlpt -v -z --port=33444 rsync.wwpdb.org::ftp/data/structures/divided/mmCIF `pwd`

but database creation failed here, too.

Just a bit of additional information: I tried breaking up the files into smaller sets based on the first character of the file name (all files 1*.cif, 2*.cif, etc.) and running with 8 threads instead of 16. The first two sets worked perfectly (creating and searching) but the third set (all the 3*.cif) failed in the createdb step. I have made available this set of files and the database that outputs prior to the segfault at this link.

Thank you again.

[martin-steinegger]

Thank you for sharing the data and putting so much effort into it. I found that models with DNA/RNA in the ATOM records cause issues. This might explain why it crashes. I will look into it how to remove them.

[martin-steinegger]

Okay I pushed one chain length fix but there is some more cleaning todo in createdb

[cjneely10]

Thank you so much for looking into this,

I was able to create a database for the entire PDB cif set and to search it successfully.

Thanks again!