build database using protein sequences?

[gancao]

How can I build my own database for HHblits? First, how can I installed self-compiled HH-suite3 version includes MPI support? If this failed, could I just use "cstranslate -i %s -o %s -A %s/cs219.lib -D %s/context_data.lib -x 0.3 -c 4 -I ca3m -b" to build?

[milot-mirdita]

You need to have OpenMPI and it's development files installed on your system. Once all of these are present HH-suite will automatically pick them up and compile the _mpi suffix binaries. Please ask your local cluster admins for advice with MPI as that is usually very specific to the local environment.

For a equivalent single node execution to cstranslate_mpi also pass the --ffindex flag to cstranslate so it does not try to read single files but full ffindex databases.