Add query to ref coord mapping

[chilampoon]

targeting spliced read issue #118 (comment)

[chilampoon]

The dumb version of this function is:

def get_ref_coords(ops, lens):
    ref_coords = []
    s = 1

    for op, l in zip(ops, lens):
        if op in [3]:
            s += l
        elif op in [1,4,5]:
            continue
        elif op in [0,2,7,8]:
            # mapped
            pos_list = list(range(s, s + l))
            ref_coords.extend(pos_list)
            s += l
        
    return ref_coords

[marcus1487]

I still feel like there is extra logic here. I think the fix here might be as simple as completely dropping the skip cigar ops. Do you get the same results if you switch the REF_OPS skip code value to False without any other code changes?

[chilampoon]

I was thinking about the same thing I think you're right - just changing the intron operation to False in REF_OPS can work as well... got the same results for genomic reads and didn't raise errors for transcriptomic reads

[chilampoon]

I put the same reference coordinate function to io.py now because it's a way to get a ref coordinate to query mapping with the exact same length as read.ref_to_signal. The get_reference_positions() from pysam doesn't include deletion bases, whereas its get_reference_sequence() includes deletions...
I tested the gapless reads without introns with updated code and got the same outputs, seems like there're many other issues for intronic reads, I'll stick to gapless read analysis for a while

[marcus1487]

I'm not sure I understand the issue. The get_reference_positions function is never used in Remora. Why is support for this function required in Remora?

I'll stick to gapless read analysis for a while

Are you saying that I should close this PR? Or would you like for this PR to be considered still?