nanoporetech
threshold of extract call of m6A
Hello developer!!
I ran the extract call for m6a modification analysis, but I don know how to set the threshold of Probability and base quality. is there are some standards for filtiering m6A call??
in some issues, is said using sample prob command will be useful, so I ran it. these are the result file. but still I do not know how to set the threshold...
Can you anvise me about this..?
Hello @Seongmin-Jang-1165,
Sorry about the delay, and thanks for sending this data over.
Taking a look at the proportions of m6A in your runs (in this case barcode 2):
A large amount of the mass is down on the left ~50%. Looks like you're using the m6A/A model, so 50% is a low-confidence call (essentially a coin toss). When you see a lot of low confidence mass, it usually indicates that the sample has a very small amount of the modified base, so most of the calls are ones where the model is not sure which state to call.
I suggest using a threshold that cuts off that mass.
You can use this number for both m6A and A, so the argument would be --filter-threshold 0.9, for example.
