Combines sequences with less than N hamming distance using counts of each sequence.

merge_seqs.py

def hamdist(str1, str2):
    """Count the # of differences between equal length strings str1 and str2"""
    diffs = 0
    for ch1, ch2 in zip(str1, str2):
        if ch1 != ch2:
            diffs += 1
    return diffs

def remove_duplicates_round(df,hamm_thres=4,merge_counts=False):
    seqs = list(df.Seq.values)
    counts = list(df.Counts.values)
    c = 0
    while c<(len(counts)-1):
        if(hamdist(seqs[c],seqs[c+1]))<hamm_thres:
            if(counts[c]>counts[c+1]):
                if(merge_counts):
                    counts[c]+=counts[c+1]
                del counts[c+1],seqs[c+1]
            else:
                if(merge_counts):
                    counts[c+1]+=counts[c]
                del counts[c],seqs[c]
        else:
            c+=1
    return pd.DataFrame({'Seq':seqs,'Counts':counts})

def remove_all_duplicates(sequences,counts,hamming_thresh=5,merge_counts=False):

    """ Removes potential sequence duplicates with < hamming_thesh separation.
    If merge counts is selected, the counts from the duplicate sequences with less
    counts are added to the counts of the corresponding sequence. Otherwise, the 
    sequence with less counts is discarded with its counts.
    
    Inputs:
    sequnces - array or list of DNA sequences
    counts - array or list of counts of each sequence
    
    Outputs:
    df - pandas dataframe with a column of all unique seqs and counts.
    """
    
    df = pd.DataFrame({'Seq':sequences,'Counts':counts})
    seq_len = sequences[0]
    for i in range(seq_len):
        df = df.ix[(df.Seq.str.slice(seq_len-i)+df.Seq.str.slice(i)).order().index]
        df = remove_duplicates_round(df,hamm_thres=hamming_thresh,merge_counts=merge_counts)
        print i,
    return df