Empty Chimeric SAM Files Created By Default

[DarioS]

All chimeric alignments SAM files are empty.

$ ls -lht *Chimeric*sam | head
-rwxrwx--- 1 biostat biostat 0 Sep 11 18:01 LSRNAN9hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 17:43 LSRNAN8hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 17:22 LSRNAN7hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 17:00 LSRNAN6hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 16:41 LSRNAN5hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 16:20 LSRNAN4hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 15:57 LSRNAN3hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 15:43 LSRNAN2hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 15:30 LSRNAN10hg38Chimeric.out.sam
-rwxrwx--- 1 biostat biostat 0 Sep 11 15:08 LSRNAN1hg38Chimeric.out.sam

It is because I am using the default, which outputs Chimeric.out.junction files with content instead. It would be good if STAR didn't output blank SAM files unnecessarily (or output both files with content, to allow easy IGV viewing), by default. Version 2.7.2b.

Also, the user guide states "... --chimOutType WithinBAM is strongly recommended." but that's incompatible with STAR-Fusion. I wouldn't recommend it unless exclusively analysing gene abundance.

[alexdobin]

Hi Dario

Hi Dario,

I have checked 2.7.2d and 2.7.3a and they both do not produce any Chimeric.out.sam files.
Please try one of those versions.

STAR-Fusion indeed requires --chimOutType Junctions which outputs chimeric junctions.
For SAM-like output --chimOutType WithinBAM is recommended over --chimOutType SeparateSAMold.
You can also use them both --chimOutType Junctions WithinBAM in one run.

Cheers
Alex

[DarioS]

You can also use them both --chimOutType Junctions WithinBAM in one run.

This is quite rare for command-line programs. Could the recommendation in the user guide be reworded in that case? Currently, it is "... --chimOutType WithinBAM is strongly recommended." But, often the biologist collaborator asks about gene fusions a few weeks after all of the data is processed which means that it needs realignment of the entire data set to get the junctions file for STAR-Fusion's processing.

[alexdobin]

Hi Dario,

I will try to reword it - you can also do it and submit a pull request.
Note that STAR-Fusion has a set of recommended parameters, so you may want to think in advance how to merge them with your parameters.

Cheers
Alex

[crutching]

@alexdobin I see that trying to run --chimMultimapNmax with --chimOutType Junctions WithinBAM does not however work. I'm torn on how to deal with this, since I'd like to provide both output types for analysis.

[alexdobin]

Hi @jhl667

there is a PR #802 that implements the multimapping chimeras output to BAM. I am going to pull it in in the next 1-2 weeks, hopefully. In the meantime you can alpha-test it if you wish. :)

Cheers
Alex

[DarioS]

I am going to pull it in in the next 1-2 weeks, hopefully.

Is there a new estimated time of availability for the desirable functionality?