variant record normalization

variant_record.py

class VariantRecord(object):
    def __init__(self, chrom, pos, ref, alt,
                 ref_genome=ref_genome):
        self.chrom = chrom
        self.pos = int(pos)
        self.ref = ref
        self.alt = alt
        if isinstance(ref_genome, RefGenome):
            self.ref_genome = ref_genome
        else:
            self.ref_genome = RefGenome(ref_genome)
        self.id = id
    
    @staticmethod
    def is_empty(allele):
        if allele is None or allele == '-' or allele == '.' or allele == '':
            return True
        return False

    def is_single(self):
        if len(self.ref) == 1 or len(self.alt) == 1:
            return True
        return False

    def fill_empty(self):
        if self.is_empty(self.ref):
            self.pos -= 1
            prev_base = self.ref_genome.base(self.chrom, self.pos)
            self.ref = prev_base
            self.alt = prev_base + self.alt
        if self.is_empty(self.alt):
            self.pos -= 1
            prev_base = self.ref_genome.base(self.chrom, self.pos)
            self.ref = prev_base + self.ref
            self.alt = prev_base

    def left_alignment(self):
        while self.alt[-1] == self.ref[-1]:
            self.pos -= 1
            prev_base = self.ref_genome.base(self.chrom, self.pos)

            def slide(allele):
                return prev_base + allele[:-1]

            self.ref = slide(self.ref)
            self.alt = slide(self.alt)

    def parsimony(self, strand='left'):
        if strand == 'right':
            while self.ref[-1] == self.alt[-1] and not self.is_single():
                self.ref = self.ref[:-1]
                self.alt = self.alt[:-1]
        if strand == 'left':
            while self.ref[0] == self.alt[0] and not self.is_single():
                self.ref = self.ref[1:]
                self.alt = self.alt[1:]
                self.pos += 1

    def normalization(self):
        if self.style not in ['snp', 'ref']:
            self.fill_empty()
            self.left_alignment()
            self.parsimony()