will need a little follow up then
uploaded a built version to this branch in case it helps...sometimes hard to build custom releases GMOD/jbrowse#1585
Tried copying files from commits GMOD/jbrowse@a372758 and GMOD/jbrowse@aeb0b02. Neither worked with forward slashes (did not test them without forward slash).
Also tried installing JBrowse-1.16.12-alpha.0.zip. Ran setup script. Fails to display sequences (both with and without forward slashes).
To create a custom build or setup from the dev branch u may need to use a GitHub clone
Basically can create a build from the dev branch like this; git clone https://github.com/GMOD/jbrowse; cd jbrowse-components; git checkout dev; ./setup.sh
Then setup.sh runs we pack and stuff (can use yarn build or yarn watch to just run we pack instead of setup.sh too)
Webpack*
And then maybe an un-noted bit of info is that you can then copy that whole GitHub clone directory wherever is needed, similar to however the zip file you downloaded from the pr. It takes a special setup to create the smaller bundles like I uploaded to the pr (have to run a makefile plus some perl stuff...can advise if that is needed too)
The built versions don't use sec/JBrowse it is all webpacked and minified
That's why basically ya need the whole GitHub repo just to make a patch. Kinda annoying but kinda just what the webpack world has brought on us
Hello, I am wondering if anyone has tried anything like the following. We have a VCF track, where the file has 1000s of samples and a lot of annotations. When rendering the browser view, most of that information and parsing is simply not used/needed. We only actually do anything with genotypes when the details view is opened. Further, when viewing a large region this track lags, which probably is related to a lot of download/parsing (but have not formally profiled this). I am considering making a plugin that would allow this track to query a slimmed down VCF (no genotypes, less INFO annotation) as the primary data source. The details view would need to be hooked into, and provide this with an alternate VCF URL. When a single variant is clicked, we'd query one position from this alternate VCF and parse/load that information. Has anyone attempted something like this before?
which jbrowse version are you using?
jb1, in some cases, tries to not spend time parsing all the genotypes
in jb2 it actually parses all genotypes by default, which means in some cases it can actually be somewhat slower than jb1 for vcf with very long genotype lists
so it will actually benefit a lot from such a system
we have tested loading 1000 genomes type data in jb2 and its not too bad but something like you are proposing would be interesting
i actually made a fun silent movie browsing 1000 genome trio dataset here the other day :) https://www.youtube.com/watch?v=zy3nC4OfP2U
if you have sample data you'd like us to look at we can check it out...see if this type of thing is needed
it may end up for that particular thing a custom detail panel that performs a fetch against a databse may be the best bet, as it may be hard to generalize
but we can surely find something to work :)
Interesting buggy behavior; I'll file a bug report unless there is an obivous fix. When I load this link: https://wormbase.org/tools/genome/jbrowse-simple/?data=data%2Fc_elegans_PRJNA13758&loc=V%3A14373361..14373473&tracks=Curated_Genes%2CDNA&highlight= the refernce sequence is misaligned, as that gene glyph marks a start of translation, but the ATG isn’t aligning to the start of the glyph. If I close and reload the reference sequence track, it now shows the ATG start at the right place. Any idea what’s going on?
And, I should note, this behavior is reproducible on my Mac in Chrome, Firefox and Safari.
maybe make an issue for that..
Yeah, there is more than one track in trackList.json with the key ""Reference sequence (DNA)”. The code that builds trackList is indiscriminately pulling in configs that start with “c_elegans” but because we have three assemblies, it’s pullng in the ref seq tracks for all three, so it appears to be a (so far predictable) crap shoot as to which ref seq you get. Lovely.
scottcain @scottcain goes off to rebuild, test, redeploy in shame.