def rep_correlation(coords1, coords2):
dists1 = misc.distsFromCoords(coords1)
dists2 = misc.distsFromCoords(coords2)
assert dists1.shape == dists2.shape
#convert to vectors
dists1 = dists1.flatten()
dists2 = dists2.flatten()
#remove zeroes
indices = np.where(dists1 != 0)[0]
dists1 = dists1[indices]
dists2 = dists2[indices]
assert len(np.where(dists2 == 0)[0]) == 0
r, p = st.pearsonr(dists1, dists2)
return r
import numpy as np
import sys
sys.path.append("..")
import data_tools as dt
import array_tools as at
import misc
#"true" distance matrix
cluster = dt.clusterFromBed(bedpath, None, None)
contactMat = dt.matFromBed(bedpath, cluster)
distMat = at.contactToDist(contactMat)
at.makeSymmetric(distMat)
for j in range(len(distMat)): #remove diagonal
distMat[j, j] = 0
chromthreed_distMat = misc.distsFromCoords(
"Chromosome3D/output/chr22_100kb/chr22_100kb_coords.tsv")
chromthreed_r = misc.pearson(distMat, chromthreed_distMat)
mmds_distMat = dt.clusterFromFile(
"hic_data/GM12878_combined_22_10kb_mmds_coords.tsv").distMat()
mmds_r = misc.pearson(distMat, mmds_distMat)
cmds_distMat = dt.clusterFromFile(
"hic_data/GM12878_combined_22_10kb_cmds_coords.tsv").distMat()
cmds_r = misc.pearson(distMat, cmds_distMat)
minimds_distMat = dt.clusterFromFile(
"hic_data/GM12878_combined_22_10kb_minimds_coords.tsv").distMat()
minimds_r = misc.pearson(distMat, minimds_distMat)
mogen_distMat = misc.distsFromCoords(
def rep_correlation(coords1, coords2):
dists1 = misc.distsFromCoords(coords1)
dists2 = misc.distsFromCoords(coords2)
return misc.pearson(dists1, dists2)
cmds_rs[i] = misc.pearson(cmds_true_mat, cmds_distMat)
minimds_structure = dt.structure_from_file(
"hic_data/GM12878_combined_{}_10kb_minimds_coords.tsv".format(chrom))
contactMat = dt.matFromBed(bedpath, minimds_structure)
minimds_true_mat = at.contactToDist(contactMat)
at.makeSymmetric(minimds_true_mat)
for j in range(len(minimds_true_mat)): #remove diagonal
minimds_true_mat[j, j] = 0
minimds_distMat = misc.distMat(minimds_structure)
minimds_rs[i] = misc.pearson(minimds_true_mat, minimds_distMat)
mogen_coords = np.loadtxt(
"MOGEN/examples/hiC/output/GM12878_combined_{}_10kb_rep1_coords.tsv".
format(chrom))
mogen_distMat = misc.distsFromCoords(mogen_coords)
mogen_rs[i] = misc.pearson(
mmds_true_mat,
mogen_distMat) #mMDS and MOGEN use the same matrix input procedure
chrom_sizes = np.loadtxt("chrom_sizes_10kb.txt")
fig = plt.figure()
ax = fig.add_subplot(111, frameon=False)
ax.plot(chrom_sizes,
mmds_rs,
linestyle="None",
marker="o",
markerfacecolor="r",
mec="r",
markersize=10,
contactMat = dt.matFromBed(bedpath, cluster)
distMat = at.contactToDist(contactMat)
at.makeSymmetric(distMat)
for j in range(len(distMat)): #remove diagonal
distMat[j,j] = 0
mmds_distMat = dt.clusterFromFile("hic_data/GM12878_combined_{}_10kb_mmds_coords.tsv".format(chrom)).distMat()
mmds_rs[i] = misc.pearson(distMat, mmds_distMat)
cmds_distMat = dt.clusterFromFile("hic_data/GM12878_combined_{}_10kb_cmds_coords.tsv".format(chrom)).distMat()
cmds_rs[i] = misc.pearson(distMat, cmds_distMat)
minimds_distMat = dt.clusterFromFile("hic_data/GM12878_combined_{}_10kb_minimds_coords.tsv".format(chrom)).distMat()
minimds_rs[i] = misc.pearson(distMat, minimds_distMat)
mogen_distMat = misc.distsFromCoords("MOGEN/examples/hiC/output/GM12878_combined_{}_10kb_rep1_coords.tsv".format(chrom))
mogen_rs[i] = misc.pearson(distMat, mogen_distMat)
chrom_sizes = np.loadtxt("chrom_sizes_10kb.txt")
fig = plt.figure()
ax = fig.add_subplot(111, frameon=False)
ax.plot(chrom_sizes, mmds_rs, linestyle="None", marker="o", markerfacecolor="r", mec="r", markersize=10, label="Standard metric MDS")
ax.plot(chrom_sizes, cmds_rs, linestyle="None", marker="o", markerfacecolor="g", mec="g", markersize=10, label="Classical MDS")
ax.plot(chrom_sizes, minimds_rs, linestyle="None", marker="o",markerfacecolor="b", mec="b", markersize=10, label="miniMDS")
ax.plot(chrom_sizes, mogen_rs, linestyle="None", marker="o",markerfacecolor="m", mec="m", markersize=10, label="MOGEN")
x_offset = 1000 #small number to prevent things from getting cut off
y_offset = 0.01
xmin = min(chrom_sizes) - x_offset
xmax = max(chrom_sizes) + x_offset
ymin = 0 - y_offset