why single end reads align to the reference genome reversely

[kcm.eid]

Hi guys,
I ve 76bp long RNAseq single end reads. I ve used Tophat to align these reads onto human genome sequence. After aligning the reads, I browsed the alignment file using samtools tview. As the reads are single end, they should align with the reference genome only in forward direction. Most of the reads are in forwad direction. But there are many lowercase reads, I suppose they are the reads aligned to the reference in reverse direction.

I am confused what should I do. I think these reads have aligned to the reference in reverse direction due to sequence homology (inverted repeats). I will use the BAM file for SNP calling in my downstream analysis. I m not confident over thse alignments. What should I do? Should I discard them? Is there any tool to discard such reads?

Thanks.

[mastal]

Unless the library-prep was done using a stranded protocol, reads will align to both the forward and reverse strand.

However, as you mentioned, reads that are in lower case may well represent reads that are aligned to repeats.

[dpryan]

Reads will align in both the forward and reverse orientation regardless. The strandedness of the library-prep will affect the likelihood of seeing anti-sense reads. BTW, I don't think tophat ever produces "lower-case" alignments (there's only one case in BAM files and that's what tophat produces). If the sequence in your reference genome is lower case, then yes that's probably indicative of a soft-masked repeat, though there are a LOT of those so I wouldn't read too much into it.

Perhaps you can post a few example reads that you're concerned about and we can then assess if there's actually anything to worry about.

[Chipper]

view uses lowercase for reverse reads. Since genes are on both strands this is expected even for stranded sequencing.