bug fixes

common_tads.py

@@ -15,6 +15,7 @@
 from tools.common_domains import find_common_domains, common_domains_multiple_sets, save_domains_matrix
 from os import path
 import glob
+from tools.str2bool import str2bool
 
 
 def main():
@@ -28,7 +29,7 @@ def main():
     parser.add_argument("-o", "--output",
                         help="Directory to save output file. Output is saved only when analysing multiple sets of TADs "
                              "(When --bedfile_1 is a directory. If None save in input directory.", default=None)
-    parser.add_argument("-r", "--report", help="If True print output matrix to stdout. Default=True", default=True, type=bool)
+    parser.add_argument("-r", "--report", help="If True print output matrix to stdout. Default=True", default=True, type=str2bool)
     parser.add_argument("-s", "--shift", default=0, type=int,
                         help="Accepted shift of two domain boundaries positions in base pair.")
     args = parser.parse_args()

moc.py

@@ -9,6 +9,7 @@
     save_moc_matrix, add_row_and_columns_id
 from os import path
 import glob
+from tools.str2bool import str2bool
 
 
 def main():
@@ -23,7 +24,7 @@ def main():
     parser.add_argument("-o", "--output",
                         help="Directory to save output file. Output is saved only when analysing multiple sets of TADs "
                              "(When --bedfile_1 is a directory. If None save in input directory.", default=None)
-    parser.add_argument("-r", "--report", help="If True return MoC to stdout. Default=True", default=True)
+    parser.add_argument("-r", "--report", help="If True return MoC to stdout. Default=True", default=True, type=str2bool)
     args = parser.parse_args()
 
     if args.output:

plot_venn.py

@@ -8,6 +8,7 @@
 import glob
 from tools.measure_of_concordance import read_domains_from_bedfile
 from tools.common_domains import plot_venn_diagram_of_3_sets, plot_venn_diagram_of_2_sets
+from tools.str2bool import str2bool
 
 
 def main():
@@ -16,7 +17,7 @@ def main():
                         help='List of files (two or three) with different domains sets.', required=True)
     parser.add_argument("-o", "--output",
                         help="Directory or filename to save a plot in.", default=None)
-    parser.add_argument("-s", "--show", help="If True show the plot.", default=True, type=bool)
+    parser.add_argument("-s", "--show", help="If True show the plot.", default=True, type=str2bool)
     args = parser.parse_args()
 
     if args.output:

tools/common_domains.py

@@ -14,38 +14,21 @@ def find_common_domains(set1, set2, shift):
     common_domains = []
     for i in range(len(set1)):
         for j in range(len(set2)):
-            if shift == 0:
-                if check_domains(domain1=set1[i], domain2=set2[j]):
-                    common_domains.append(set1[i])
-            else:
-                if check_shifted_domains(domain1=set1[i], domain2=set2[j], shift=shift):
-                    common_domains.append(set1[i])
+            if check_shifted_domains(domain1=set1[i], domain2=set2[j], shift=shift):
+                common_domains.append(set1[i])
     return common_domains
 
 
 def check_shifted_domains(domain1, domain2, shift):
     """Check if two domains positions are identical (with accepted shift)"""
     if domain1[0] != domain2[0]:
         return False
-    if domain2[1] - shift < 0:
-        start = 0
-    else:
-        start = domain2[1] - shift
-    if domain1[1] in range(start, domain2[1] + shift):
-        if domain1[2] in range(domain2[2] - shift, domain2[2] + shift):
+    if domain2[1] - shift <= domain1[1] <= domain2[1] + shift:
+        if domain2[2] - shift <= domain1[2] <= domain2[2] + shift:
             return True
     return False
 
 
-def check_domains(domain1, domain2):
-    """Check if two domains have exact same boundaries"""
-    if domain2[0] != domain1[0]:
-        return False
-    if domain1[1] == domain2[1] and domain1[2] == domain2[2]:
-        return True
-    return False
-
-
 def save_domains_matrix(tad_matrix, outfile):
     """Format nicely and save conserved domains matrix with sets names"""
     tad_matrix.to_csv(outfile, sep=",", header=True, index=True)
@@ -56,10 +39,11 @@ def common_domains_multiple_sets(domains_sets, shift):
     """Return a matrix with common domains for different sets."""
     matr = np.zeros((len(domains_sets), len(domains_sets)), dtype=int)
     for i in range(len(domains_sets)):
-        for j in range(len(domains_sets)):
+        for j in range(i, len(domains_sets)):
             if i == j:
-                continue
-            matr[i][j] = len(find_common_domains(set1=domains_sets[i], set2=domains_sets[j], shift=shift))
+                matr[i][j] = len(domains_sets[i])
+            else:
+                matr[i][j] = matr[j][i] = len(find_common_domains(set1=domains_sets[i], set2=domains_sets[j], shift=shift))
     return matr
 
 

tools/measure_of_concordance.py

@@ -23,24 +23,27 @@ def read_domains_from_bedfile(bedfile):
 def calculate_moc(set1, set2):
     n1 = len(set1)  # number fo domains in set 1
     n2 = len(set2)  # number of domains in set 2
-    if set1 == set2 == 1:
+    if n1 == n2 == 1:
         return 1
     moc = 0
     for i in range(n1):
         p_i = set1[i][2] - set1[i][1]  # domain length
         for j in range(n2):
             if set1[i][0] == set2[j][0]:  # check if same chromosome
                 q_j = set2[j][2] - set2[j][1]  # domain length
-                overlap = len(check_overlap(start1=set1[i][1], end1=set1[i][2], start2=set2[j][1],
-                                            end2=set2[j][2]))  # calculate overlap in bp
+                overlap = check_overlap(start1=set1[i][1], end1=set1[i][2], start2=set2[j][1],
+                                            end2=set2[j][2])  # calculate overlap in bp
                 if overlap > 0:
                     moc += (overlap ** 2 / (p_i * q_j))
     return (1 / ((n1 * n2) ** (1 / 2) - 1)) * (moc - 1)
 
 
 def check_overlap(start1, end1, start2, end2):
     """Return the overlap between two domains"""
-    return range(max(start1, start2), min(end1, end2 + 1))
+    overlap = min(end1, end2) - max(start1, start2)
+    if overlap < 0:
+        overlap = 0
+    return overlap
 
 
 def moc_for_multiple_sets(set_list):

tools/str2bool.py

@@ -0,0 +1,12 @@
+import argparse
+
+def str2bool(v):
+    if isinstance(v, bool):
+        return v
+    v = v.lower()
+    if v in ('true', 't', '1'):
+        return True
+    elif v in ('false', 'f', '0'):
+        return False
+    else:
+        raise argparse.ArgumentTypeError()