Sequence Viewer context menu "blast" should provide more options

[Elaine Meng]

... instead of always just using defaults, which prevents blasting alphafold database (for example). Should just open Blast Protein tool but is currently blocked by that tool only allowing structure chains as the query sequence (ticket #5055)

[Zach Pearson]

This is fixed I think - the sequence viewer will now open Blast Protein panels from its right-click menu.

[Elaine Meng]

In UCSF ChimeraX version: 1.6.dev202212122321 (2022-12-12), using Sequence Viewer right-click menu to open Blast Protein gives a traceback:

Traceback (most recent call last):

File "/Users/meng/Desktop/ChimeraX_Daily.app/Contents/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/chimerax/seq_view/tool.py", line 634, in <lambda>

lambda: BlastProteinTool?(self.session, sequences = StringArg?.unparse("%s:1" % self.alignment.ident))

File "/Users/meng/Desktop/ChimeraX_Daily.app/Contents/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/chimerax/blastprotein/ui/tool.py", line 71, in init

self._build_ui()

File "/Users/meng/Desktop/ChimeraX_Daily.app/Contents/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/chimerax/blastprotein/ui/tool.py", line 187, in _build_ui

self.menu_widgetsuniprot_or_seq_input?.setText(self._sequences)

KeyError?: 'uniprot_or_seq_input'

KeyError?: 'uniprot_or_seq_input'

File "/Users/meng/Desktop/ChimeraX_Daily.app/Contents/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/chimerax/blastprotein/ui/tool.py", line 187, in _build_ui
self.menu_widgetsuniprot_or_seq_input?.setText(self._sequences)

See log for complete Python traceback.

[Zach Pearson]

​Sorry, I removed a vestigial reference and it started working. Now I can right click and hit blast protein, and the sequence is pre-filled.

[Elaine Meng]

(A) Is this supposed to populate the Blast Protein tool with the "raw sequence" of the chosen entry?  Or is it supposed to put in a sequence specifier?   By sequence specifier I mean what is documented here

name-of-alignment:N 

where N specifies the Nth sequence in the alignment.  II do get something like the sequence specifier, but Query type is set to "Raw Sequence."  If I used it on a single sequence shown with

open 2gbp
seq chain /A

Then the field contains

1/A:1

...which is not raw sequence.  If it is really supposed to be a sequence specifier, the Query type menu should include an entry for that type, maybe "Sequence Specifier" or "From Viewer".  I tried just applying the search with the above and it gave results, but they were totally different and with lower scores than if I search directly by setting the query to chain A of the 2gbp structure.

(B) Another issue occurs with a multiple sequence alignment, e.g. super8.msf (attached). Using the context menu to open Blast Protein for a specific sequence, no matter which one I choose, the "Raw Sequence" field is populated with

super8.msf:8

...the last sequence since there are 8 total,  even if chose any of the other (non-last) sequences from the context menu.
 

super8.msf​

[Elaine Meng]

Added by email2trac

[pett]

I made it put in the correct sequence-alignment index. I could supply the raw sequence instead if preferred.

[pett]

I tried to reproduce the "chain vs. sequence search yields different results" and the results were the same for me. I guess try again?

[Elaine Meng]

In today's build, I do get exactly the same search results from using the chain sequence directly from structure and from using it from the Seequence Viewer.  So either I messed up before, or something changed.  Either way, thumbs up!

When starting Blast Protein from the sequence viewer, I would vote for putting the "raw sequence" instead of the sequence identifier into the input dialog, since the field is labeled Raw Sequence.

[pett]

Okay, it fills in the actual sequence instead of the alignment index now. I _think_ that covers everything, so closing this ticket. If I forgot something, let me know.

[Elaine Meng]

Looks good: calling Blast Protein from Sequence Viewer context menu now fills in the "Raw Sequence" field with the amino acid sequence of the chosen one.  Thanks!