abPOA fails to capture difference in prefix

Hi,

Thank you for developing abPOA!

I am trying to use abPOA to build consensus and cluster short TR sequences (~40-70bp in length). I've noticed that if the difference in sequences is in prefix, then abPOA identifies only one cluster instead of two.

I generated three different dataset of 20 reads using this function:

def generate_subreads(num, where = "start"):
    seqs = []
    repeat = "CTAT"
    insertion="TCTA"
    if where == "start":
        for i in range(num//2):
            seqs.append(SeqRecord(Seq(insertion + repeat*15), id = str(i), name= str(i), description=""))
    elif where == "middle":
        for i in range(num//2):
            seqs.append(SeqRecord(Seq(repeat*8 + insertion + repeat*7), id = str(i), name= str(i), description=""))
    else:
        for i in range(num//2):
            seqs.append(SeqRecord(Seq(repeat*15 + insertion), id = str(i), name= str(i), description=""))

    for i in range(num//2):
        seqs.append(SeqRecord(Seq(repeat*15), id = str(num//2 + i), name= str(num//2 + i), description=""))

    return seqs

and run: aligner.msa(seqs, out_cons=True, out_msa=True, max_n_cons=2, min_freq=0.3)

and here is the output of the aligner.msa on these datasets: synthetic_output.txt

Is there a problem with abpoa algorithm on such corner case?

Thank you, Tatiana

yangao07 / abPOA

abPOA fails to capture difference in prefix #57