samtools index: failed to create index

[juancresc]

I got this error while trying with a large genome (triticum aestivum).

Required bamfile index iwgsc_mites/merged_alignments.bam.bai not found. Attempting to create with samtools index.
[E::hts_idx_push] Region 536871232..536871256 cannot be stored in a bai index. Try using a csi index with min_shift = 14, n_lvls >= 6
samtools index: failed to create index for "iwgsc_mites/merged_alignments.bam": Numerical result out of range
FAILED - Aborting.

What I did as a workaround was to add -c to samtools index command (line 2703) system "samtools index -c $$options{'bamfile'}";

and change the expected index name (line 2697) by changing .bai to .csi at the end $expected_idx = "$$options{'bamfile'}" . ".csi";

I'm not sure whether this will have other consequences, would be nice to have an option to use bai or csi.

[MikeAxtell]

Thanks Juan for pointing that out. Until your message I wasn't actually aware that there was a .csi type of bam index; I guess I've never encountered an assembly with very large contigs (according to this post on biostars the max contig size that .bai indices can handle is 2^29-1 bases).

I can't say for sure or not whether the changes you've made would have unintended consequences. I agree this option should be fixed and I'll try to get to that next time I do an update.

Can you update on this thread if you do test out the changes you made, and how they worked?

Thanks, Mike

[juancresc]

Thank you for your answer. Seems like it did run as expected (at least no errors).

The only thing I'm noticing (and might be unrelated) is when I specify a locifile, the valids miRNAs found are quite different as those without. I would expect those miRNA found with locifile are contained in the results without locifile. What am I missing here?

[MikeAxtell]

Hi Juan, thanks for following up. Good to know when I implement a fix. It might not be until this summer because of my crazy schedule right now.

about your locifile miRNA question: The exact coordinates used by the locifile were different than those found in de-novo run, I'm guessing. ShortStack's MIRNA calling is quite dependent upon the exact coordinates it used for folding possible RNAs. This can cause false negatives when the de novo cluster-calling produces sub-optimal coordinates. I suspect this is the answer to your question.

On Mon, Feb 4, 2019 at 2:50 PM Juan Manuel Crescente < notifications@github.com> wrote:

Thank you for your answer. Seems like it did run as expected (at least no errors).

The only thing I'm noticing (and might be unrelated) is when I specify a locifile, the valids miRNAs found are quite different as those without. I would expect those miRNA found with locifile are contained in the results without locifile. What am I missing here?

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FMikeAxtell%2FShortStack%2Fissues%2F88%23issuecomment-460387552&data=02%7C01%7Cmja18%40psu.edu%7C5ccd754621a3491465ec08d68ada1113%7C7cf48d453ddb4389a9c1c115526eb52e%7C0%7C0%7C636849066523264076&sdata=%2FzGLR8nD6TmBeqR3v7YhRg251rRpmvrEKl6C2Vjwo1A%3D&reserved=0, or mute the thread https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAGiXiVLEGr95O5n8uXEGB-g1ra9c8lB3ks5vKI8XgaJpZM4aeWUt&data=02%7C01%7Cmja18%40psu.edu%7C5ccd754621a3491465ec08d68ada1113%7C7cf48d453ddb4389a9c1c115526eb52e%7C0%7C0%7C636849066523274085&sdata=9zijnUfP%2FauGky1m78U%2BF%2B4Sk%2F71KUZzvpPm2xiJOjM%3D&reserved=0 .

-- Michael J. Axtell, Ph.D. Professor of Biology The Pennsylvania State University http://sites.psu.edu/axtell http://plantsmallrnagenes.science.psu.edu

[juancresc]

I'll keep testing, thanks!