Define behaviour for genome "unknown"

[brainstorm]

If the lab reports an "unknown" genome, bcbio-nextgen crashes because there's no genome defined with that name.

A reasonable default would be to not do any alignment in that case.

[vals]

Did we reach a decision on this? The minimal test set seem to have "unknown".

@b97pla For testing, I just added a line in bowtie_indices.loc and sam_fa_indices.loc that is a copy of the first line but with the name changed to "unknown".

[brainstorm]

Added "unknown" to the scripts that generate those .loc files. We need to mantain idempotency in the reference genome downloading/generation.

chapmanb/cloudbiolinux#27

[b97pla]

In the long run, I'd vote for the unaligned-bam-file-creation approach I
mentioned above. I'll try to have a look at this next week.

On Mon, Feb 13, 2012 at 5:14 PM, Roman Valls <
[email protected]

wrote:

Added "unknown" to the scripts that generate those .loc files. We need to
mantain idempotency in the reference genome downloading/generation.

chapmanb/cloudbiolinux#27

---

Reply to this email directly or view it on GitHub:
#78 (comment)

---

Pontus Larsson, PhD
Science For Life Laboratory
www.scilifelab.se

Email: [email protected]
Phone: +46 8 5248 1440

Mobile: +46 76 946 9788

[vals]

Reopening this issue in the light of @b97pla 's comment above.

[brainstorm]

According to Brad:

I'm definitely working on better non-human support for the pipeline. I
don't have plans for unknown/assembly genomes right now. I think the
best bet there would be to funnel into a separate pipeline that
handles assembly.

@vezzi , maybe something worth keeping in mind?