Milot Mirdita

How did you do the searches? How did you create the `swissprot_human` subset? I am surprised you are getting results at all for your test 2). The default parameters of...

Please try to reverse the search direction (sequences vs profiles, not profiles vs sequences). It looks like the small number of queries is causing some weird issue.

I remember `blastdbcmd` having issues, however, I don't remember what was wrong. We use a different workflow to assign taxids for the NR: https://github.com/soedinglab/MMseqs2/blob/804bb2af6d1be4086252c46bf15f3c75a5d9e931/data/workflow/databases.sh#L419 The download part for the `accession2taxid`...

This looks like it installed a really old version of MMseqs2. Did you install the software through conda into a new environment?

Seems like something is wrong with the installed compilers within this conda environment. You should be able to install the `compilers` conda meta package which should contain everything to compile...

This is possible but a bit tricky: Please make one FASTA file containing all sequence entries. Then call `createdb` ``` mmseqs createdb YOUR_FASTA_FILE.fasta db ``` Then take a look at...

Oh, also this will only work for protein sequences. Nucleotide sequences need a diagonal seed point to compute the alignment, which would be much more difficult to hack.

You can try. The tsv you need to create has the same format of, but with two more columns score (can be 0) and match diagonal (position i-j): ``` query_key...

You can ignore the score, its not used further in the `align` module. Just set it to 0. You can pass only one diagonal per query-target pair. You should be...

We don't implement any clustering like that, as its usually biologically not very meaningful. You can compute a sparse all-vs-all search and cluster based on scores with whatever clustering algorithm...