y_means = []
z_means = []
x_lengths = []
y_lengths = []
z_lengths = []
with open(design_file) as infile:
for line in infile:
cell_type1, cell_type2 = line.strip().split()
for chrom in chroms:
path1 = "hic_data/{}_{}_{}kb.bed".format(cell_type1, chrom, res_kb)
path2 = "hic_data/{}_{}_{}kb.bed".format(cell_type2, chrom, res_kb)
if os.path.isfile(path1) and os.path.isfile(path2):
structure1, structure2 = multimds.full_mds(path1,
path2,
penalty=penalty)
structure1.rescale()
structure2.rescale()
r, t = la.getTransformation(structure1, structure2)
structure1.transform(r, t)
#compartments
contacts1 = dt.matFromBed(path1, structure=structure1)
contacts2 = dt.matFromBed(path2, structure=structure2)
compartments1 = np.array(
ca.get_compartments(contacts1, structure1))
compartments2 = np.array(
ca.get_compartments(contacts2, structure2))
from multimds import compartment_analysis as ca
from multimds import data_tools as dt
from scipy import stats as st
from matplotlib import pyplot as plt
import numpy as np
from multimds import linear_algebra as la
from scipy import signal as sg
from multimds import multimds as mm
path1 = "hic_data/GM12878_combined_19_100kb.bed"
path2 = "hic_data/K562_19_100kb.bed"
struct1, struct2 = mm.full_mds(path1, path2, prefix="test_")
mat1 = dt.matFromBed(
"hic_data/GM12878_combined_{}_{}kb.bed".format(chrom, res_kb), struct1)
comps1 = ca.get_compartments(mat1, struct1)
mat2 = dt.matFromBed("hic_data/K562_{}_{}kb.bed".format(chrom, res_kb),
struct2)
comps2 = ca.get_compartments(mat2, struct2)
r, p = st.pearsonr(comps1, comps2)
if r < 0:
comps1 = -comps1
comp_diffs = np.abs(comps1 - comps2)
dists = np.array([
la.calcDistance(coord1, coord2)
for coord1, coord2 in zip(struct1.getCoords(), struct2.getCoords())
])
from multimds import multimds
import sys
chrom = sys.argv[1]
multimds.full_mds("hic_data/GM12878_combined_{}_100kb.bed".format(chrom), "hic_data/K562_{}_100kb.bed".format(chrom))
from multimds import multimds
import sys
iteration = sys.argv[1]
chrom = sys.argv[2]
strain = sys.argv[3]
multimds.full_mds("hic_data/ctrl_{}_{}_32kb.bed".format(strain, chrom),
"hic_data/galactose_{}_{}_32kb.bed".format(strain, chrom),
weight=0,
penalty=0.1,
prefix=iteration + "_")
import numpy as np
from multimds import plotting as plot
from multimds import multimds as mm
mappability = np.loadtxt("mappability_21_5kb.bed", usecols=3)
#struct1 = dt.structure_from_file("GM12878_combined_21_5kb_structure.tsv")
#struct2 = dt.structure_from_file("K562_21_5kb_structure.tsv")
struct1, struct2 = mm.full_mds("hic_data/GM12878_combined_21_5kb.bed",
"hic_data/K562_21_5kb.bed")
mappability = mappability[
struct1.chrom.minPos / struct1.chrom.res +
struct1.nonzero_abs_indices()] #only loci in structures
mappable = np.where(mappability > 0.75)
struct1.points = struct1.points[mappable]
struct2.points = struct2.points[mappable]
plot.plot_structures_interactive((struct1, struct2))
from multimds import multimds
import sys
chrom = sys.argv[1]
strain = sys.argv[2]
multimds.full_mds("hic_data/ctrl_{}_{}_32kb.bed".format(strain, chrom),
"hic_data/galactose_{}_{}_32kb.bed".format(strain, chrom),
weight=0,
penalty=0.1)
import sys
from multimds import multimds
celltype1 = sys.argv[1]
celltype2 = sys.argv[2]
multimds.full_mds("hic_data/{}_21_100kb.bed".format(celltype1),
"hic_data/{}_21_100kb.bed".format(celltype2))
from multimds import plotting as plot
from multimds import multimds as mm
struct1, struct2 = mm.full_mds("ctrl_Scer_12_32kb.bed",
"galactose_Scer_12_32kb.bed",
weight=0,
penalty=0.1)
plot.plot_structures_interactive((struct1, struct2), out_path="sup6a.png")
struct1, struct2 = mm.full_mds("ctrl_Scer_12-upstream_32kb.bed",
"galactose_Scer_12-upstream_32kb.bed",
weight=0,
penalty=0.1)
plot.plot_structures_interactive((struct1, struct2),
out_path="sup6b_upstream.png")
struct1, struct2 = mm.full_mds("ctrl_Scer_12-downstream_32kb.bed",
"galactose_Scer_12-downstream_32kb.bed",
weight=0,
penalty=0.1)
plot.plot_structures_interactive((struct1, struct2),
out_path="sup6b_downstream.png")
path1 = sys.argv[1]
path2 = sys.argv[2]
prefix1 = os.path.basename(path1.split(".")[0])
prefix2 = os.path.basename(path2.split(".")[0])
n = 10
all_r_sq = []
ps = np.arange(0, 0.1, 0.01)
for p in ps:
all_changes = []
for i in range(n):
multimds.full_mds(path1, path2, penalty=p)
structure1 = dt.structure_from_file("{}_structure.tsv".format(os.path.basename(prefix1)))
structure2 = dt.structure_from_file("{}_structure.tsv".format(os.path.basename(prefix2)))
if p == 0:
r, t = la.getTransformation(structure1, structure2)
structure1.transform(r,t)
all_changes.append(np.array([la.calcDistance(coord1, coord2) for coord1, coord2 in zip(structure1.getCoords(), structure2.getCoords())]))
r_sq = []
for i in range(n):
for j in range(i):
r, p = st.pearsonr(all_changes[i], all_changes[j])
r_sq.append(r**2)
import matplotlib
from matplotlib import pyplot as plt
from multimds import multimds
from multimds import linear_algebra as la
struct1, struct2 = multimds.full_mds("sim1_chr21_100kb.bed", "sim2_chr21_100kb.bed")
gen_coords = struct1.getGenCoords()
dists = [la.calcDistance(coord1, coord2) for coord1, coord2 in zip(struct1.getCoords(), struct2.getCoords())]
plt.subplot2grid((10,10), (0,0), 9, 10, frameon=False)
plt.plot(gen_coords, dists, lw=2)
#define offsets
xmin = min(gen_coords)
xmax = max(gen_coords)
x_range = xmax - xmin
x_start = xmin - x_range/25.
x_end = xmax + x_range/25.
ymin = 0
ymax = max(dists)
y_range = ymax - ymin
y_start = ymin - y_range/25.
y_end = ymax + y_range/25.
#define axes with offsets
plt.axis([x_start, x_end, y_start, y_end], frameon=False)
#plot axes (black with line width of 4)
plt.axvline(x=x_start, color="k", lw=4)
res_kb = 100
prefix1 = "GM12878_combined"
prefix2 = "K562"
path1 = "hic_data/{}_{}_{}kb.bed".format(prefix1, chrom, res_kb)
path2 = "hic_data/{}_{}_{}kb.bed".format(prefix2, chrom, res_kb)
size1 = dt.size_from_bed(path1)
size2 = dt.size_from_bed(path2)
ps = np.arange(0, 0.6, 0.1)
errors = np.zeros_like(ps)
for i, p in enumerate(ps):
#os.system("python ../multimds.py -P {} {} {}".format(p, path1, path2))
structure1, structure2 = mm.full_mds(path1, path2, penalty=p)
#structure1 = dt.structure_from_file("{}_{}_{}kb_structure.tsv".format(prefix1, chrom, res_kb))
#structure2 = dt.structure_from_file("{}_{}_{}kb_structure.tsv".format(prefix2, chrom, res_kb))
dists1 = dt.distmat(path1, structure1, size1)
dists2 = dt.distmat(path2, structure2, size2)
errors[i] = np.mean(
(error(dists1,
structure1.getCoords()), error(dists1, structure1.getCoords())))
xs = ps
x_int_size = 0.1
ys = errors
y_int_size = 0.05
x_start = min(xs) - x_int_size / 4.
from multimds import plotting as plot
from multimds import multimds as mm
struct1, struct2 = mm.full_mds("hic_data/GM12878_combined_21_10kb.bed", "hic_data/K562_21_10kb.bed", weight=0)
plot.plot_structures_interactive((struct1, struct2))
from multimds import data_tools as dt
import numpy as np
from multimds import compartment_analysis as ca
from sklearn import svm
from multimds import linear_algebra as la
from mayavi import mlab
from multimds import multimds as mm
path1 = "hic_data/GM12878_combined_21_100kb.bed"
path2 = "hic_data/K562_21_100kb.bed"
struct1, struct2 = mm.full_mds(path1, path2)
contacts1 = dt.matFromBed(path1, struct1)
enrichments1 = np.loadtxt("binding_data/GM12878_21_100kb_active_coverage.bed",
usecols=6)
bin_nums1 = struct1.nonzero_abs_indices() + int(
struct1.chrom.minPos / struct1.chrom.res)
enrichments1 = enrichments1[bin_nums1]
comps1 = np.array(ca.get_compartments(contacts1, struct1, enrichments1))
contacts2 = dt.matFromBed(path2, struct2)
enrichments2 = np.loadtxt("binding_data/K562_21_100kb_active_coverage.bed",
usecols=6)
bin_nums2 = struct2.nonzero_abs_indices() + int(
struct2.chrom.minPos / struct2.chrom.res)
enrichments2 = enrichments2[bin_nums2]
comps2 = np.array(ca.get_compartments(contacts2, struct2, enrichments2))
coords1 = struct1.getCoords()
coords2 = struct2.getCoords()