Possibility to include NAs in taxonomy csv format for easier parsing?

[LanderDC]

Hi,

I'm trying to parse the taxonomy column of the _virus_summary.tsv file, but unclassified taxonomic levels are omitted which causes inconsistencies. In the table below for example, for the first entry, Straboviridae would end up in the Order column while it is a viral family.

seq_name	length	topology	coordinates	n_genes	genetic_code	virus_score	fdr	n_hallmarks	marker_enrichment	taxonomy
NODE_A75_length_5000_cov_17.505180_BlackFly28	5000	No_terminal_repeats	NA	4	11	0.9997	0.0003	1	6.8731	Viruses;Duplodnaviria;Heunggongvirae;Uroviricota;Caudoviricetes;Straboviridae
NODE_A20_length_7615_cov_6.791589_BlackFly6	7615	No_terminal_repeats	NA	11	11	0.9997	0.0003	6	15.3910	Viruses;Duplodnaviria;Heunggongvirae;Uroviricota;Caudoviricetes
NODE_A8_length_6639_cov_9.592350_BlackFly34	6639	No_terminal_repeats	NA	8	11	0.9997	0.0003	1	13.4767	Viruses;Duplodnaviria;Heunggongvirae;Uroviricota;Caudoviricetes
NODE_A4_length_12404_cov_730.807901_BlackFly44	12404	No_terminal_repeats	NA	6	11	0.9996	0.0004	2	3.4366	Viruses;Riboviria;Orthornavirae;Negarnaviricota;Monjiviricetes;Mononegavirales;Rhabdoviridae

Would it be possible to output the taxonomy column as: Viruses;Duplodnaviria;Heunggongvirae;Uroviricota;Caudoviricetes;;Straboviridae;; or Viruses;Duplodnaviria;Heunggongvirae;Uroviricota;Caudoviricetes;unclassified;Straboviridae?

Thanks in advance!

[apcamargo]

Hi @LanderDC,

You make a good point. I'm not sure about adding an "unclassified" taxon to the lineage (your second option) because it might confuse users. They could think that geNomad couldn't assign the sequence to an order but could assign it to a family, when in reality, the family isn't part of any order at all

I like your first option, though.

Another possibility would be adding prefixes to the taxa names, like GTDB does? For instance:

Viruses;r__Duplodnaviria;k__Heunggongvirae;p__Uroviricota;c__Caudoviricetes;f__Straboviridae;

This could make parsing easier. The problem is that "Viruses" is not a taxon at all and doesn't have a named rank.

[LanderDC]

Thanks for your swift reply!

You are right that my second option might cause some confusion. I like your suggestion to add prefixes, because you immediately see which taxonomic rank you're dealing with and don't have to remember which suffix belongs to which rank. However, if possible I would still like empty places for ranks that are not yet classified by ICTV, because then you can just split the taxonomy column based on the semicolon:

Viruses;r__Duplodnaviria;k__Heunggongvirae;p__Uroviricota;c__Caudoviricetes;;f__Straboviridae;;

About the "Viruses" problem, I don't feel that it is really an issue but I understand that you would like to have everything consistent.

[apcamargo]

I like your suggestion. The "Viruses" thing does bother me, though. If I add the additional semicolons, the ranks would be could be inferred by the number of the column, so the prefixes are not strictly required.

[LanderDC]

True, whatever is easier to implement should be fine 🙂

[apcamargo]

I'll will work on that for the next release :)

[LanderDC]

Nice, thanks!

[LanderDC]

Are you ok with a pull request on this issue? I forked your repo and slightly modified taxonomy.py to get these empty fields for unclassified ranks (https://github.com/LanderDC/genomad/commit/14eb85d458ebecdd28fda021765cdbad46e9214e). Incidentally, the "Viruses" rank is also removed by these modifications.

[apcamargo]

Thanks, @LanderDC

If it's easy enough, I'd appreciate that. I'd need to evaluate what could break with this change (the removal of the "Viruses" taxon), but I don't forsee any issues.

[LanderDC]

Great!

[apcamargo]

Fixed in version 1.8.0. Thanks @LanderDC!