Haplotype phasing – HapCUT
The idea was neat, thanks to the stronger assumption (but still realistic) to have several fragments. The transformation into max-cut problem is nice. I missed the elimination of homozygote alleles, so did not understood the paper in the beginning. In fact this should be a standard trick for this problem, since it results in a stronger assumption.
Question raised: can we compute the probability that 2 adjacent SNP contain a recombination sites between them? If yes, it helps to merge subsequent SNPs into an allele, reducing the guessing one has to make during phasing
Gene/genome alignment with shuffling/inversion taken into consideration.
Prof Ben’s paper is nice. However, I wondered why I didn’t come up with this idea. Certainly, I did not have exp with such kind of alignment where the evolutionary distance is large enough for shuffling/inversion to appear.
– Should we generalize the edit distance to contain inversion & duplication (with weight assigned). In this case, only closely related sequences should be compared, as otherwise the distance does not reflect much of the evolutionary distance?
– How does Brujin graph resolve the repeated genomic regions?