Using End-to-end alignment results in zero chimeric alignments?

[Katherine Pillman]

Hello,

Main point: When using the default alignEndsType, I 

Back story:

I am relatively new to STAR and have been using it in conjunction with CIRCexplorer to identify circular RNAs. 

I noticed in STAR was aligning softclipping some reads that were mapped across chimeric junctions by another program. 

I found a post that suggested using end-to-end mapping might fix this (https://groups.google.com/forum/#!topic/rna-star/XZgOmZe9Tpg).

I performed exactly the same star run as previously but using the --alignEndsType EndToEnd parameter. 

The output has 0 chimeric alignments (where the identical run without EndToEnd identified 497114 chimeric alignments).

Can anyone shed any light on why this might be? Is this expected behaviour - i.e. should EndToEnd not be used for detecting chimeras?

Thanks!

Katherine

[Alexander Dobin]

Hi Katherine,

the EndToEnd option should *not* be used for chimeric detection, as it will force all the chimerically split reads to align locally.

What parameters are you using for chimeric detection?

Cheers

Alex

[Katherine Pillman]

Hi Alex, thanks for your response!

My command for the non-end-to-end chimeric detection run is:

/data/sacgf/tools/STAR/STAR-STAR_2.5.0c/bin/Linux_x86_64_static/STAR --genomeDir /data/sacgf/reference/iGenomes/Homo_sapiens/UCSC/hg19/Sequence/STAR150Index/ \
        --runThreadN 8 \
        --runMode alignReads \
        --readFilesIn  <one_single_end_fastq_file.gz>\
        --readFilesCommand zcat \
        --chimSegmentMin 20 \
        --twopassMode None \
        --outSAMtype BAM SortedByCoordinate \
        --quantMode GeneCounts \
        --outSAMunmapped Within KeepPairs \

Is this what you wanted?

Cheers, Katherine

[Alexander Dobin]

Hi Katherine,

the command looks good. 

You could also play with other parameters. For instance, STAR-Fusion uses:

--chimSegmentMin 12 --chimJunctionOverhangMin 12 --alignSJDBoverhangMin 10 --alignMatesGapMax 200000 --alignIntronMax 200000 --chimSegmentReadGapMax 3 --alignSJstitchMismatchNmax 5 -1 5 5  --seedSearchStartLmax 20 --winAnchorMultimapNmax 50

Cheers

Alex

[Katherine Pillman]

Thanks Alex, I will try adjusting some of those parameters instead.

Cheers, Katherine