Load from Server with custom-appended genome

[Scott Kallgren]

I work regularly with a reference (human, mouse, rat, other) genome appended with a couple custom chromosomes that don't exist in the genome. I have a number of goals/problems with this, one of which is to easily load in standard annotations and tracks such as I can when I click "File > Load from Server..." while using an IGV-built-in genome ("Human hg19", for instance). When I try to Load from Server using a custom local genome, even if it is the Ensembl or Gencode standard, Load from Server fails.

Sample error message: "No datasets are available for the current genome (~/Ensembl/Homo_sapiens.GRCh38.dna.primary_assembly.fa)."

Is there a way for me to get this to work? By renaming this FASTA genome file? Creating a .genome named "Human hg19" instead? Downloading whatever IGV uses internally for "Human hg19" and appending chromosomes to that? Some option in the XML session file?

I'm running IGV 2.4.3 on macOS Sierra 10.12.6.

Thanks much for any assistance.
Scott

---

P.S. In case others have opinions, one other question I'm trying to solve is:

Is it possible for me to dynamically concatenate custom contigs to a standard genome FASTA as a pseudo-file (or served 'file') on a server only when requested? Many of our experiments have 1-2 short custom contigs added, and it is inefficient to create and store a 99.9%-identical 3GB human reference FASTA for each. Indexing (.fai) could be a problem here, possibly. The same applies to a GTF or BED feature file: I would like to, ideally, concatenate the extra sequences upon request. I'm sure I'll have to write some custom code for this, but am just looking for pointers as to how to think about it and get it to work most easily. Ideally, IGV would take 2+ FASTA files as an input, but I don't see a way to do this unless I use a directly to create a .genome, which I can't script.

I'm happy to make this latter question a separate post if advised.

[James Robinson]

For the short term you could try this hack.   Load the standard IGV genome,  then close igv.   

Locate the ".genome" file in the igv/genomes folder and move it to another directory (move don't copy,  we don't want to leave the standard hg19.genome file in the genomes folder)

Unzip the .genome file  (its a standard zip file),  for example 

    unzip hg19.genome

Edit the file "property.txt" in a plain text editor,  change the "sequenceLocation" field to point to your fasta

Remove the hg19.genome file,  then zip everything else up

   zip hg19_modified.genome *

Load the hg19_modified.genome from the Genomes > Load from file menu

This should work, but has not been verified.  The standard annotations are in the .genome zip file,  load-from-server works off the genome ID in the property.txt file.

--

---
You received this message because you are subscribed to the Google Groups "igv-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to igv-help+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/igv-help/11f58eec-80d7-4765-bb7c-32d13e358744%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Scott Kallgren]

Thanks, Jim. Following your instructions, I got this to work using local copies of the Gencode genome (.fa and .fa.gz) in place of the S3 hg19.fasta sequenceLocation and compressedSequence in properties.txt. IGV took more time (10 seconds) after I requested Load from Server than it did with the default hg19 (instantly?), FYI.

This is a quick workaround but won't allow me to point to a bunch of different appended versions of the human genome that, say, I have on S3 or locally, right? Ideally I would want my users to be able to choose the appropriate reference genome variation from a list of several appended versions in their IGV session, but still be able to use Load from Server to load additional tracks.

Thanks again,

Scott

To unsubscribe from this group and stop receiving emails from it, send an email to igv-help+u...@googlegroups.com.

[James Robinson]

I suggest you open a git issue with details summarized of what you want to do.   I doubt we will support appending to genomes, but could support some sort of alias genome ID so you can still use "load from server".   WRT timing of the menu appear,  that is just random network latency,   there is no difference in the request from your customized or the standard genome.