Tweet
Hi,
I have a bizarre problem with what looks like read errors that occur predominantly at the end of reverse orientation reads (i.e. at the read start on forward orientation). My reads are 101bp single end RNASeq reads from a large number of different barley samples. I have quality trimmed the reads from each sample using the standard cutoff of 20 and mapped them to my reference (full length cDNA sequences) with Bowtie1, then deduplicated each mapping and merged them all into a single BAM file.
I have attached a Tablet screenshot showing what's going on. I have looked through a large subset of the different reference sequences in my mapping and it's very obvious that the read errors are concentrated at the end of the reverse reads, like in the screenshot (blue = reverse, green = forward). There is no equivalent of this in the forward orientation reads.
The base qualities of the mismatched bases in the reverse reads are all in the range of 30-35, i.e. they are supposedly good base calls, so there is no problem with the trimming here. A FASTQC image of the base qualities is also attached, and it does look like the read quality at the read start is relatively poor, but it's still well in the green zone. Also, if this was the problem, it should affect forward orientation reads too.
Could this be a base call calibration problem? If it was, then why would it only affect the reverse reads?
thanks
Micha
I have a bizarre problem with what looks like read errors that occur predominantly at the end of reverse orientation reads (i.e. at the read start on forward orientation). My reads are 101bp single end RNASeq reads from a large number of different barley samples. I have quality trimmed the reads from each sample using the standard cutoff of 20 and mapped them to my reference (full length cDNA sequences) with Bowtie1, then deduplicated each mapping and merged them all into a single BAM file.
I have attached a Tablet screenshot showing what's going on. I have looked through a large subset of the different reference sequences in my mapping and it's very obvious that the read errors are concentrated at the end of the reverse reads, like in the screenshot (blue = reverse, green = forward). There is no equivalent of this in the forward orientation reads.
The base qualities of the mismatched bases in the reverse reads are all in the range of 30-35, i.e. they are supposedly good base calls, so there is no problem with the trimming here. A FASTQC image of the base qualities is also attached, and it does look like the read quality at the read start is relatively poor, but it's still well in the green zone. Also, if this was the problem, it should affect forward orientation reads too.
Could this be a base call calibration problem? If it was, then why would it only affect the reverse reads?
thanks
Micha