STAR unmapped: too short

[jk1124]

Hello,

I recently mapped 8 RNA-Seq samples using STAR. The samples are paired. The data in log.final.out for each sample look very similar to the sample's partner - with one exception.

I have one sample that shows 6.57% reads mapped to multiple loci and 32.38 % of reads unmapped: too short. It's partner shows 24.30% and 13.12%, respectively (these numbers more closely resemble the other samples). Uniquely mapped reads: roughly 60% for both.

I am wondering why the first sample has such a high percentage of "unmapped:too short" reads. I've read that sequencing quality can affect this number, so I expected the the sample to be low quality. However, Fastqc shows that the sequencing length and base quality is very good (and similar) for both samples. Can anyone help me to explain this?

Thanks for your help in advance!

[GenoMax]

Did you scan these samples with a trimming program (for adapters, adapter dimers, read-through etc) before you did the alignments? It is possible that quality/adapter contamination differs between the two. Until trimmed sequence length will always be identical for all reads (with the exception of those samples run on MiSeq and analyzed by MiSeq reporter on MiSeq).

[jk1124]

Thanks for your reply! I thought that if either of the samples were to have more adapter contamination, it would be the first sample (with a high number of reads unmapped:too short). Is my thinking correct?

I tried trimming the reads for this sample but saw similar numbers when I mapped:

Uniquely mapped reads % | 60.98%
% of reads mapped to multiple loci | 6.61%
% of reads unmapped: too short | 31.76%

This is the command I used:

java -jar Trimmomatic-0.33/trimmomatic-0.33.jar PE -phred33 R1.fastq.gz R2.fastq.gz R1_paired.fq.gz R1_unpaired_fq.gz R2_paired.fq.gz R2_unpaired.fq.gz ILLUMINACLIP:Trimmomatic-0.33/adapters/TruSeq3-PE-2.fa:2:30:10

What do you think?

[GenoMax]

Grab a few of those unmapped reads and blast them against nt to see if you have chimeras or some contamination that you were not expecting (since they survived trimming but are not mapping with STAR).