Hi everybody,
I have a question regarding my MeDip-seq data (single end, Illumina, meC and hmeC AB from Diagnode). I comparing linage related cell types (neuronal) and so far everything looks alright, could quite nicely call a lot of peaks using MACS2. However after i visualize my data i notice that a lot of peaks have a huge strand bias, meaning reads for this peaks are not mixed on + or - (http://pasteboard.co/2tYMz9CZ.png , hmedip of three different linage related cell types). I know that there is the possibility of CpN methylation or that only one strand is methylated but I was thinking this are rather rare event however I have a lot of peaks with a strand bias (have to find a way to quantify but would say 60-70 %). Another point would be amplification artifact but even after removal of duplicated reads i still get the strand bias. Furthermore I see this effect in all my replicates (which are quite consistent: cor 0.98) and across my different cell types (which share also a quite high similarity in their methylation pattern). Is there another possibility and for my downstream analysis how should I handle such peaks (analysis both strands separately ?).
Thanks for any suggestion,
Flo
I have a question regarding my MeDip-seq data (single end, Illumina, meC and hmeC AB from Diagnode). I comparing linage related cell types (neuronal) and so far everything looks alright, could quite nicely call a lot of peaks using MACS2. However after i visualize my data i notice that a lot of peaks have a huge strand bias, meaning reads for this peaks are not mixed on + or - (http://pasteboard.co/2tYMz9CZ.png , hmedip of three different linage related cell types). I know that there is the possibility of CpN methylation or that only one strand is methylated but I was thinking this are rather rare event however I have a lot of peaks with a strand bias (have to find a way to quantify but would say 60-70 %). Another point would be amplification artifact but even after removal of duplicated reads i still get the strand bias. Furthermore I see this effect in all my replicates (which are quite consistent: cor 0.98) and across my different cell types (which share also a quite high similarity in their methylation pattern). Is there another possibility and for my downstream analysis how should I handle such peaks (analysis both strands separately ?).
Thanks for any suggestion,
Flo
Comment