search

PeteHaitch / methtuple

methtuple is a methylation caller for methylation events that co-occur on the same DNA fragment from high-throughput bisulfite sequencing data, such as methylC-seq.
MIT License
12 stars 3 forks source link

Relevant changes in Bismark v0.14.4 #99

Open PeteHaitch opened 9 years ago

PeteHaitch commented 9 years ago

From http://www.bioinformatics.babraham.ac.uk/projects/bismark/RELEASE_NOTES.txt:

Changed the FLAG values of paired-end alignments to the CTOT or CTOB strands so that reads can be properly displayed in SeqMonk when imported as BAM files. This change affects only paired-end alignments in --pbat or --non_directional mode. In detail we simply swapped the Read 1 and Read 2 FLAG values round so reads now resemble exactly concordant read pairs to the OT and OB strands. Note that results produced by the methylation extractor or further downstream of that are not affected by this change. FLAG values now look like this:

                                  Read 1       Read 2

                          OT:         99          147

                          OB:         83          163

                          CTOT:      147           99

                          CTOB:      163           83
PeteHaitch commented 8 years ago

I don't have any --pbat or --non_directional libraries with which to test. However, I think this will cause problems with methtuple. Specifically, read-1 when mapped to CTOT or CTOB has the "second in pair" flag set when it is really the first in the pair. SImilarly, read-2 when mapped to CTOT or CTOB has the "first in pair" flag set when it is really the second in the pair. So this change seems like a workaround in order to get Seqmonk to display things in a desired way, but it also breaks the meaning of the flags according to the SAM specifications. Have raised with Felix https://github.com/FelixKrueger/Bismark/issues/41