Hi Ben,

Thanks for the suggestions. The sorting by group worked pretty well.

There are some situations where alphabetical sorting would be helpful (to
get all residues of a certain type together at the first position and then
by the second position. I guess I can group on a single column but then
that doesn't sort the second column.  But I can work with the sorting by
group.

By the way, Jim Procter gave me some "Groovy" scripts to left-justify some
selected columns or right-justify some selected columns. I am finding them
really, really helpful in cleaning up alignments that are very gappy in
some loop regions and I don't care about the pairwise alignments in that
region; I can left-justify them and then delete empty columns. He said it
was an idea to include that as a regular feature. It's a little time
consuming to read in the left-script, and then the right-script, and then
the left-script again since sometimes it takes a lot of both scripts to
clean up an entire alignment in a way that makes sense. For my two cents, I
think that would be great if it were just a preloaded tool that would make
it easy to left and right justify and set of selected columns.

Thanks,
Roland



---------------------------------------------
Roland Dunbrack (he/his/him/him)
Institute for Cancer Research
Fox Chase Cancer Center
Philadelphia PA 19111
http://dunbrack.fccc.edu
http://dunbrack.org


On Sat, Jul 4, 2020 at 3:57 PM Benedict Soares (Staff) <
[email protected]> wrote:

> Hi Roland,
>
> Another possibility you might try (which for your first use case will do
> exactly what you want, and for your second use case get you most of the
> way):
>
> 1) Select the column(s) you are interested in:
>     to select a single column just click at the top of the column to
> select a single column;
>     to select multiple contiguous columns click and drag at the top of the
> columns;
>     to select multiple non-contiguous columns, select the first column(s),
> then hold the Ctrl (or cmd if on a Mac) key down whilst selecting further
> column(s).
>
> 2) Click on "Select" -> "Make Groups for Selection" which will create a
> group for each different combination of residues/gaps _in your selected
> columns_.  The groups themselves will currently be scattered throughout the
> sequences, so...
>
> 3) Click on "Calculate" -> "Sort" -> "By Group" which will then re-arrange
> the sequences so that sequences in the same group are together.  These
> groups will be ordered from largest group to smallest group(s), or the
> other way round. Select Sort By Group again to reverse the order they're in.
>
> In your first use case, you should have the sequences ordered with all the
> sequences with a gap in the selected column, followed by the single
> sequence with a residue in the column (or the other way round if you
> reverse the order).
>
> In the second use case it won't be in alphabetical order, but in order of
> size of groups (i.e. occurrences of each residue/gap combination).  You
> might find it easier to see what's going on in those groups/columns by
> selecting "View" -> "Hide" -> "All but Selected Region" (assuming you still
> have the relevant columns selected).  When you've found your sequences of
> interest, select them (click on the ID) and you can bring back the hidden
> columns with "View" -> "Show" -> "All Columns".
>
> Hope I've understood correctly and this helps,
>
> Ben
> ------------------------------
> *From:* [email protected] <
> [email protected]> on behalf of Roland Dunbrack <
> [email protected]>
> *Sent:* 04 July 2020 11:34
> *To:* [email protected] <[email protected]>
> *Subject:* [Jalview-discuss] sorting by sequence selection
>
> Hello,
>
> Is there a way to sort all the sequences or a set of them in a Jalview
> session by some selected columns? For instance, I have a gapped column in
> 1000 sequences and it's a pain to try to find the inserted residue. If I
> can just sort by that column it would move to the top or the bottom of the
> sequences. In other cases, I am looking for different mutations at some
> positions in antibodies, and if I can select 3 columns from a set of
> sequences and then reorder those whole sequences in alphabetical order of
> the selected 3-residue segment in each.
>
> Thanks,
> Roland
>
> ---------------------------------------------
> Roland Dunbrack (he/his/him/him)
> Institute for Cancer Research
> Fox Chase Cancer Center
> Philadelphia PA 19111
> http://dunbrack.fccc.edu
> http://dunbrack.org
>
> The University of Dundee is a registered Scottish Charity, No: SC015096
>
_______________________________________________
Jalview-discuss mailing list
[email protected]
http://www.compbio.dundee.ac.uk/mailman/listinfo/jalview-discuss

Reply via email to