So - last week, Lawrence Kelley and spent around 3 days hacking a
prototype link between phyre2 and Jalview. Here's the story so far:
Our aim is to provide Jalview users a link to submit sequences to
Phyre2, and also allow Phyre2 users to open their model results in
Jalview. The former requires extension of Jalview's existing REST
service interface and some subtle modifications of the Phyre2 server.
The latter (which also blocks the former), involves deciding how a set
of models derived from a set of template sequences (each with their own
sequence and dataset associations) can be shown aligned to the user's
Query sequence and the multiple sequence alignment used to discover
templates and create the models.
Things aren't quite ready for public consumption, but here's what is now
in the repository:
* phyre2 spike branch - special build for launching Jalview from a
phyre2 result page
- button is only shown when my dundee email used for input).
* Patches on the branch
- new STRUCTMODEL statement in annotation file
This allows specification of a Query sequence, a template seuqence
name, and URI's for a PDB file and pairwise alignment that provide the
model coordinates (pdb file with sequence matching query, but
coordinates from template structure), and alignment between query and
template
- various minor patches to make the prototype work ... see git log
Current procedure for viewing phyre2 results:
Phyre2 provides query.jal and query.jal.ann - the profile MSA for the
query (including indels) and an annotation file.
Loading query.jal, and dropping query.jal.jvannot should result in:
* The query sequence set as reference for alignment, and insertions hidden
* scores from psi-blast added as annotation to profile sequences (access
via Calculations->Sort->...)
* models and pairwise alignments are retrieved via STRUCTMODEL statement
i. Models are associated with query sequence
- can view all models by:
1. Select "View 3D Structure ..." after right clicking on query
sequence's sequence ID.
2. Select 'Cached structures' from the dropdown menu at top of 3d
structure dialog - this should show all the models loaded onto the
alignment.
3. pick one or all models, and hit 'View' - the models should be
displayed superimposed.
Outstanding implementation:
* template sequences need to be added to the query/profile MSA according
to the provided pairwise alignment (Jim working on this code)
* Model PDB files need to mapped to template sequences according to
provided pairwise alignment (Tochukwu - this is something you could look
at ...)
* Each model is associated with both the Query sequence AND the template
sequence added to the alignment - which is not normally the situation in
Jalview. We will need to come up with solutions (UI, datamodel
extensions, etc) that allows models to be coloured according to
properties derived from either the modelling process, the query/template
alignment, or annotation derived from the template sequences.
That's all for now.. see you next week (if not before via Skype!)
Jim
···
--
-------------------------------------------------------------------
Dr JB Procter, Jalview Coordinator, The Barton Group
Division of Computational Biology, School of Life Sciences
University of Dundee, Dundee DD1 5EH, UK.
+44 1382 388734 | www.jalview.org | www.compbio.dundee.ac.uk