I have some PE bs-seq data which is aligning very strangely: the left reads are aligning only to the - strand and the right reads only to the + strand. When I say only, I mean 95% of the aligned reads. I am using bowtie2 end to end with 0 minimum score (i.e. perfect matches only). Library was constructed so that left reads are C->T converted and right reads and G->A converted, so they are aligned to respectively converted genomes. Alignment percentage is reasonable at about 45%.
There are two problems with the alignment which suggest issues with sequencing: (1) this isn't a serious issue, but there is significant overlap between the pairs, reducing the usefulness of the PE strategy. (2) the reads which aligned to the upper strand have methylation data for contexts in the bottom strand. I.e. there are Cs in the aligned reads in the upper strand where there is a CG in the sequence in the lower strand.
This is all done with an established pipeline which I use regularly for SE/PE data.
When aligning with bismark, reads only align to the bottom strand.
I am not sure exactly what the issue is here, but seeing as I haven't had this issue with any other paired ends data, I am guessing something is wrong with it.
There are two problems with the alignment which suggest issues with sequencing: (1) this isn't a serious issue, but there is significant overlap between the pairs, reducing the usefulness of the PE strategy. (2) the reads which aligned to the upper strand have methylation data for contexts in the bottom strand. I.e. there are Cs in the aligned reads in the upper strand where there is a CG in the sequence in the lower strand.
This is all done with an established pipeline which I use regularly for SE/PE data.
When aligning with bismark, reads only align to the bottom strand.
I am not sure exactly what the issue is here, but seeing as I haven't had this issue with any other paired ends data, I am guessing something is wrong with it.