Filter on certain annotation values

[rvosa]

It would be a good enhancement if it was possible to filter sequences on certain annotation values that users might take as indicators of sequence quality. For example, the following sequence has an annotation that gives the identifier of the voucher specimen, which for systematists is useful information in assessing whether the taxonomic identification is actually credible: http://www.ncbi.nlm.nih.gov/nuccore/JQ922076.1 (on line 5 of the features table).

This would require that we actually populate and use the features table (https://github.com/naturalis/supersmart/blob/master/lib/Bio/Phylo/PhyLoTA/DAO/Result/Feature.pm), where the sequence should have a primary_tag whose value is specimen_voucher.

[rvosa]

I think there's a number of different issues that come into play here, none of which are superimportant but they can lead to nags from users and reviewers:

• some sequences have poor quality taxonomic annotations. Basically, they're assigned entirely (or somewhat) the wrong taxon because people are unsure about what their specimen is. One symptom of this is that these sequences are sometimes given "taxon names" that consist of Genus sp. or Genus sp.VOUCHER. It would make sense to allow these names to be filtered out when expanding clades using smrt taxize so that only names that look like decent binomials are kept.
• also, some other sequences are simply bad for lab reasons (e.g. many NN's, contaminations, etc.). It makes sense to have some sort of blacklisting facility like we have in the HTS-barcode-checker, and I think there is also something like this in PHLAWD: just allow users to throw out specific accession numbers.

On the other hand, we can also say that users need to exercise due diligence and inspect their species.tsv and their alignments.