This is indeed a somewhat odd ratio of strands, it reminds me very much of the single-cell BS-Seq technique which undergoes several rounds of PBAT-type pull-down and strand-regeneration. I have not heard of an RRBS-method that would be doing this, do you happen to know how these libraries were generated (e.g. using the Zymo Pico Methyl-Seq kit?).
I would probably just try to align only Read 1 (also in --non-directional mode) just to see if the mapping efficiency goes up a bit because 38% seems rather low).
In principle it is fine to merge files at the BAM level and then feed the merged file to the methylation extractor. A word of caution is in order for paired-end end files though if you are using Samtools merge because this merging does not normally guarantee that Read 1 and Read 2 are kept in the same order in the following file unless you specify the option -n:
Similarly, already merged files my be sorted by read name using
to bring them back into the correct format.
If the R1/R2 order is muddled with then the overlap detection and the strand assignments would go a little mad so this is to be avoided.
I generally tend to merge files right from the start at the FastQ file level because this also reduces the number of files generated in the process considerably. Hope this helps, Felix
I would probably just try to align only Read 1 (also in --non-directional mode) just to see if the mapping efficiency goes up a bit because 38% seems rather low).
In principle it is fine to merge files at the BAM level and then feed the merged file to the methylation extractor. A word of caution is in order for paired-end end files though if you are using Samtools merge because this merging does not normally guarantee that Read 1 and Read 2 are kept in the same order in the following file unless you specify the option -n:
Code:
Options: -n sort by read names
Code:
samtools sort -n
If the R1/R2 order is muddled with then the overlap detection and the strand assignments would go a little mad so this is to be avoided.
I generally tend to merge files right from the start at the FastQ file level because this also reduces the number of files generated in the process considerably. Hope this helps, Felix
Comment