cfe-lab / cfeintact Goto Github PK

Currently used structures like SeqRecord have type any. This can cause runtime type errors.

Graphical output can help users to understand HIVIntact's decisions, and call alternative judgments.

Currently an MSD error looks like this:

Query sequence has a mutated splice donor site, AT.

It would be nice to have more information here, such as the positions of discovered MSD and PSI sites, the nucletide/aminocid sequences around them, and maybe the expected vs found values. Something useful like that.

Instead of doing global alignment on whole genome sequences, align just the searched subtype ORF sequence to the full query sequence.

The searched subtype ORF sequence should be prefixed and suffixed with "+" to discourage spread of nucleotides.

It's relatively straightforward to detect mutated start/stop codons.

And it's useful to note that such mutations happened as the underlying reason for the shortness of an ORF.

It it possible that, as in case of HXB2, other subtypes include defects too.

Not all of them are significant, but frameshifts and start/stop codons misplacements impact the final verdict very much.

Currently, if a working_dir contains spaces in its name, all calls to mafft will fail.

context

It would be nice to annotate detected errors with two additional pieces of info: Severity and Confidence.

See bandit for an example.

Currently, almost all ORFs are considered intact if their size is within -30/+30 range of their subtype ORF's length.

But, as we know, some ORFs, such as env have a larger variance, and some, such as tat do not have any variance at all.

Therefore it is better to base the cut-offs on empirical data, rather than to stick to arbitrary measures.

This should provide an overview on how well different regions, and perhaps the whole sequence, align to the subtype sequence.