Re: [Gmod-ajax] JBrowse callback functions used for track color

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Hi Vaneet,

I think I've at least gotten close. There are a few tricks: you can specify
a list of transcript types that are noncoding, and you can specify what the
noncoding transcripts are colored with a callback, so I have this snippet
of a config:

         "noncodingType" :
["transcript","lnc_RNA","miRNA","pseudogene","rRNA","snRNA","snoRNA","tRNA"],
         "style" : {
            "className" : "feature",
            "unprocessedTranscriptColor" : "function(feature) {var tag =
feature.get('tag');if (tag == 'pseudoduplicate') {return 'magenta';} if
(tag == 'non-coding' ) {return 'red'}; return 'blue';}",
            "color" : "function(feature) {console.log(feature); var tag =
feature.get('tag');if (tag == 'pseudoduplicate') {return 'magenta';} if
(tag == 'non-coding' ) {return 'red'}; return 'goldenrod';}"
         },

I'm not positive you wanted the 'transcript' transcripts to be coding or
not, so I included them in noncoding. Also, I noticed that sometimes mRNAs
have tag attributes too, so I wrote the callback for both color and
unprocessedTranscriptColor.

Anyway, this should get you close to what you're looking for.

Scott

On Mon, Sep 25, 2023 at 11:31 AM Vaneet Lotay <van...@uc...>
wrote:

> Hey Scott,
>
>
>
> You can access my GFF3 files here:
>
>
>
> https://bigfrog.xenbase.org/pub/xenbase/genomics/share/
>
>
>
> So the ‘transcript-tagged’ GFF3 is where I added a ‘tag’ attribute to the 9
> th column at the transcript level as you were saying these callback
> functions usually access features at the transcript level.
>
>
>
> When that wasn’t working I changed the values in the source column (2nd
> column) to hold my tags which is what the ‘source-tagged’ GFF3 has.  This
> is for the echinoderm S. purpuratus v5.0 genome.  Do you need the genome
> sequence file as well?
>
>
>
> Thanks for looking into this, appreciate it!
>
>
>
> Vaneet
>
>
>
> *From:* Scott Cain <sc...@sc...>
> *Sent:* September 21, 2023 2:03 PM
> *To:* Vaneet Lotay <van...@uc...>
> *Cc:* gmo...@li...
> *Subject:* Re: JBrowse callback functions used for track color
>
>
>
> [△EXTERNAL]
>
>
>
> Hi Vaneet,
>
>
>
> Sorry this is being such a pain. Can you send me a url to your GFF? I'll
> try setting up a test instance to see if I can sort it out.
>
>
>
> Scott
>
>
>
>
>
> On Thu, Sep 21, 2023 at 10:27 AM Vaneet Lotay <van...@uc...>
> wrote:
>
> I still can’t get it to work properly.  Since you had mentioned that
> JBrowse extracts the feature at the transcript level I tagged these genes
> at the transcript as well as the parent gene level in the GFF3.  Also I
> decided to add the tag in the ‘source’ column in the GFF3 as that had
> worked for me in the past and perhaps easier to extract:
>
>
>
> customColor = function(feature) {
>
>         var src = feature.get('source');
>
>         if (src == 'PS-DUP') {
>
>                 return 'magenta';
>
>         } else if (src == 'Non-coding'){
>
>                 return 'red';
>
>         }
>
>         return 'goldenrod';
>
>         }
>
>
>
> However this still doesn’t work.  I tried printing to the console the
> value of the ‘src’ variable to see if it’s picking up something and it does
> at least for one of the two conditions I have (PS-DUP), so not perfect I
> suppose.  Maybe I’m missing some conditional code somewhere in the JBrowse
> source code that keep overwriting something as it seems to add red either
> to the main color or the ‘clip colour’ to the gene models.  That’s why I
> simply picked red as one of my colors to give in to this phenomenon since
> it seems like it’s targeting non-coding transcripts anyways.  Is this
> functionality any easier in JBrowse2?  I’m beginning to setup a version of
> JBrowse2 slowly, which is why I ask.
>
>
>
> Coloring gene model tracks has always been a tricky thing for me, any help
> anyone can provide would be great.
>
>
>
> Thanks,
>
>
> Vaneet
>
>
>
>
>
> *From:* Scott Cain <sc...@sc...>
> *Sent:* September 13, 2023 5:19 PM
> *To:* Vaneet Lotay <van...@uc...>
> *Cc:* gmo...@li...
> *Subject:* Re: JBrowse callback functions used for track color
>
>
>
> [△EXTERNAL]
>
>
>
> I am reasonably sure that syntax is right; it turns out that function is
> no longer used in our site, but I found a current track that has this
> callback:
>
>
>
> "color" : "function(f) {var type =  f._parent.get('predictiontype'); if
> (type == 'tmhmm') {return 'magenta';} if (type == 'seg') {return
> 'lightseagreen';} if (type == 'signalp') {return 'aquamarine'} if (type ==
> 'ncoils') {return 'chartreuse';} if (type == 'pfam') {return
> 'lightsalmon';} return 'purple'; }"
>
>
>
> BUT!!!! This applies to features that only have a parent and a child;
> coloring genes is even more tricky. I forgot that I'd run into some
> difficulties with how JBrowse was representing genes (at least 6 or 7 years
> ago), so I ended up writing my own Gene glyph that handles coloring among
> other things:
>
>
>
>
> https://github.com/WormBase/website-genome-browsers/tree/jbrowse-production/jbrowse/jbrowse/plugins/wormbase-glyphs
>
>
>
> I don't think you'll need to do this though, but I can't remember how to
> deal with it.
>
>
>
> Colin, help :-)
>
>
>
> Scott
>
>
>
>
>
>
>
> On Wed, Sep 13, 2023 at 3:30 PM Vaneet Lotay <van...@uc...>
> wrote:
>
> I suppose I can just tag everything to the transcript as that seems to
> work.  How do you override that default behavior of JBrowse trying to
> estimate that a transcript is non-coding and colors it red?  Even when I
> can successfully change the color using my callback function, that red
> color remains for some transcripts where JBrowse is controlling it.  Even
> when I try to manually change it via 'Edit config' in a temporary session
> it doesn't change the red.
>
>
>
> Thanks,
>
>
>
> Vaneet
> ------------------------------
>
> *From:* Vaneet Lotay <van...@uc...>
> *Sent:* September 13, 2023 3:36 PM
> *To:* Scott Cain <sc...@sc...>
> *Cc:* gmo...@li... <gmo...@li...>
> *Subject:* Re: [Gmod-ajax] JBrowse callback functions used for track color
>
>
>
> [△EXTERNAL]
>
>
>
> Hey Scott,
>
>
>
> Is that the right syntax for calling the parent gene feature:
>
>
>
>
>
>
>
>
>
>
>
>
>
> *parentRainbowHeatMap = function(feature) {     var score =
> feature._parent.get('score');     var max = 0.01;     var min = 2.50;
> if(score<max) {score = max;}     if (score >min) {score = min;}     var
> blueness = 190*((score-min)/(max-min));     var hex =
> Math.round(blueness).toString(16);     if (hex.length < 2) { hex = '0' +
> hex; }     var color = '#' + hex + hex + 'DD';     return color;  }*
>
>
>
> I'm asking because it results in this error from my browser console:
>
>
>
> *Uncaught TypeError: feature._parent is undefined*
>
>
>
> I tried without the period or without the underscore, still finds it
> undefined.
>
>
>
> Thanks,
>
>
> Vaneet
> ------------------------------
>
> *From:* Scott Cain <sc...@sc...>
> *Sent:* September 13, 2023 12:02 PM
> *To:* Vaneet Lotay <van...@uc...>
> *Cc:* gmo...@li... <gmo...@li...>
> *Subject:* Re: JBrowse callback functions used for track color
>
>
>
> [△EXTERNAL]
>
>
>
> Hi Vaneet,
>
>
>
> Coloring things in gene tracks can be tricky; the feature that is being
> considered when the color is applied is (I think) the transcript-level
> feature, not the parent gene. So if you've attached an attribute to the
> parent gene that you want to use, you actually have to get the parent
> feature first. As an example, I have this function:
>
>
>
> parentRainbowHeatMap = function(feature) {
>     var score = feature._parent.get('score');
>     var max = 0.01;
>     var min = 2.50;
>     if(score<max) {score = max;}
>     if (score >min) {score = min;}
>     var blueness = 190*((score-min)/(max-min));
>     var hex = Math.round(blueness).toString(16);
>     if (hex.length < 2) { hex = '0' + hex; }
>     var color = '#' + hex + hex + 'DD';
>     return color;  }
>
>
>
> that gets the parent feature's score in order to determine the color of
> the transcript.
>
>
>
> For the red transcripts--I think the gene glyph is trying to determine if
> a transcript is noncoding and if it thinks it is, it colors it red. I'm not
> positive about that though--I pretty much never use the default color in
> gene tracks :-)
>
>
>
> Scott
>
>
>
>
>
> On Tue, Sep 12, 2023 at 1:37 PM Vaneet Lotay <van...@uc...>
> wrote:
>
> Hi Scott,
>
>
>
> If you or anyone in the JBrowse community can help me with this, I’d
> appreciate it.  I’ve used callback functions before successfully to add
> links within the ‘View details’ box for various attributes.  Now what I
> would like to do is color specific genes in my GFF3 a certain color
> depending on a certain attribute.  What I’ve done is added a ‘tag’
> attribute that would indicate whether the gene is a non-coding gene or not
> as well as other types, like so at the end of the 9th column of the GFF3:
>
>
>
>
> ID=gene-LOC115920884;Dbxref=Echinobase:ECB-GENE-23010450,GeneID:115920884;Name=LOC115920884;gbkey=Gene;gene=LOC115920884;gene_biotype=lncRNA;tag=non-coding
>
>
>
> Then I defined a callback function which would extract this attribute and
> pick the color based on the value, while keeping what I believe is the
> default JBrowse track color as the backup option (goldenrod):
>
>
>
> customColor = function(feature) {
>
>         gtag = feature.get('tag');
>
>         if (gtag == 'pseudoduplicate') {
>
>                 return 'magenta';
>
>         } else if (gtag == 'non-coding'){
>
>                 return 'green';
>
>         } else{
>
>                 return 'goldenrod';
>
>         }
>
>         return 'goldenrod';
>
>         }
>
>
>
> Then under the style stanza of the track I added the call to the function:
>
>
>
>          "style" : {
>
>             "className" : "feature",
>
>             “color” : “{customColor}”
>
>          },
>
>
>
> I’ve done this for other callback functions successfully, but perhaps
> coloring the track behaves differently.  I might also be not extracting the
> attribute properly from the GFF or my code has errors, please let me know
> if this is the case.  I have been able to extract other attributes from the
> 9th column of GFF3 before so I know it’s possible, but just not working
> as expected here.  Also it seems some of my gene models already seem to be
> colored red for some reason without this callback function even being used
> as opposed to goldenrod, I’m not sure if that’s default JBrowse behaviour,
> can you explain why that happens?
>
>
>
> Thanks,
>
>
> Vaneet
>
>
>
> *Vaneet Lotay*
>
> *Xenbase Bioinformatician*
>
> *University of Calgary*
>
> *2500 University Drive NW*
>
> *Calgary, AB T2N 1N4*
>
> *CANADA*
>
>
>
>
>
> --
>
> ------------------------------------------------------------------------
> Scott Cain, Ph. D.                                                  scott
> at scottcain dot net
>
> GMOD Project Manager (http://gmod.org/)
> 216-392-3087
>
> WormBase Developer (http://wormbase.org/)
>
> Alliance of Genome Resources Group Leader (http://alliancegenome.org/)
>
> VirusSeq Project Manager (https://virusseq-dataportal.ca/)
>
> Human Cancer Models Initiative Project Manager (
> https://hcmi-searchable-catalog.nci.nih.gov/)
>
>
>
> --
>
> ------------------------------------------------------------------------
> Scott Cain, Ph. D.                                                  scott
> at scottcain dot net
>
> GMOD Project Manager (http://gmod.org/)
> 216-392-3087
>
> WormBase Developer (http://wormbase.org/)
>
> Alliance of Genome Resources Group Leader (http://alliancegenome.org/)
>
> VirusSeq Project Manager (https://virusseq-dataportal.ca/)
>
> Human Cancer Models Initiative Project Manager (
> https://hcmi-searchable-catalog.nci.nih.gov/)
>
>
>
> --
>
> ------------------------------------------------------------------------
> Scott Cain, Ph. D.                                                  scott
> at scottcain dot net
>
> GMOD Project Manager (http://gmod.org/)
> 216-392-3087
>
> WormBase Developer (http://wormbase.org/)
>
> Alliance of Genome Resources Group Leader (http://alliancegenome.org/)
>
> VirusSeq Project Manager (https://virusseq-dataportal.ca/)
>
> Human Cancer Models Initiative Project Manager (
> https://hcmi-searchable-catalog.nci.nih.gov/)
>

-- 
------------------------------------------------------------------------
Scott Cain, Ph. D.                                                  scott
at scottcain dot net
GMOD Project Manager (http://gmod.org/)
216-392-3087
WormBase Developer (http://wormbase.org/)
Alliance of Genome Resources Group Leader (http://alliancegenome.org/)
VirusSeq Project Manager (https://virusseq-dataportal.ca/)
Human Cancer Models Initiative Project Manager (
https://hcmi-searchable-catalog.nci.nih.gov/)