I am using STAR to align paired-end sequencing data. I have reads whose mate lengths are 150bp. I have not trimmed the adapter sequences as it has been suggested by others in STAR. I have go around 85% total alignment, while the remaining was tagged as unmapped due to short lengths.
Out of curiosity I selected some reads at random and ran BLAST, manually omitting the adapter sequence from the read (retaining 110 bases from each mate). I found it completely matched to a gene with 100% identity and coverage in BLAST. I checked that both mates of the same fragment were present and they both had identical BLAST results.
Upon examining the sequences further, I found that the reads overlapped 100%, i.e, one was essentially the reverse complement of the other. Though this is unusual, I can't understand why STAR is omitting these reads from the main alignment.
Your thoughts on this?
