In [ ]:

    

from scipy.spatial.distance import hamming
import skbio
from skbio import DNA, DistanceMatrix, TabularMSA
import numpy as np


    

In [ ]:

    

aln = TabularMSA.read('all_genomes-0.001.fasta', constructor=DNA)
aln.reassign_index(minter="id")


    

In [ ]:

    

dm = DistanceMatrix.from_iterable([seq.values for seq in aln], metric=hamming, keys=aln.index)


    

In [ ]:

    

dm


    

In [ ]:

    

1/ dm['B', 'D']


    

In [ ]:

    

1/np.min(dm.condensed_form())


    

In [ ]:

    

d = DistanceMatrix.read("/home/kevin/ws/kwip-experiments/coalescent/runs/run_2016-03-22/data/9671/kwip/50x-0.001-wip.dist")


    

In [ ]:

    

d


    

In [ ]:

    

d.filter(sorted(d.ids))


    

In [ ]: