Handle unnormalized HiC data

Author: e-sollier

figeno/cli/cli.py

@@ -2,7 +2,7 @@
 
 from figeno.cli import gui, init,make
 
-__version__ = "1.3.2"
+__version__ = "1.3.3"
 
 def main():
     parser = ArgumentParser("figeno",formatter_class=ArgumentDefaultsHelpFormatter)

figeno/figeno.py

@@ -89,7 +89,7 @@ def __init__(self,config=None,config_file=None,reference=None,genes_file=None,cy
                 for s in config["regions"]:
                     if not "chr" in s: raise KnownException("Please provide at least the chromosome for each region (and optionally the start and end coordinates, otherwise figeno will assume than the region spans the whole chromosome).")
                     if s["chr"]=="" or s["chr"]=="chr" or s["chr"] is None: raise KnownException("Please provide at least the chromosome for each region (and optionally the start and end coordinates, otherwise figeno will assume than the region spans the whole chromosome).")
-                    chr = str(s["chr"]).lstrip("chr")
+                    chr = str(s["chr"]).lstrip("chr").lstrip("Chr")
                     orientation = s["orientation"] if "orientation" in s else "+"
 
                     if "start" in s and (s["start"] is not None) and (s["start"]!=""): 

figeno/gui/package.json

@@ -1,6 +1,6 @@
 {
   "name": "figeno",
-  "version": "1.3.2",
+  "version": "1.3.3",
   "private": true,
   "homepage": "./",
   "dependencies": {

figeno/track_hic.py

@@ -64,6 +64,11 @@ def draw(self, regions, box ,hmargin,warnings=[]):
         max_bindist=self.max_dist//resolution /2
         n = len(regions)
 
+        self.balanced= "weight" in c.bins()[:].columns
+        if not self.balanced:
+            warnings.append("No balancing weights were found for file "+self.file+". Figeno will show the raw, unnormalized data. If you want to show normalized data, please run: \"cooler balance "+self.file+"\" and re-generate the figure."\
+                            " See https://cooler.readthedocs.io/en/latest/cli.html#cooler-balance for more information and options.")
+
         angle = self.find_angle(regions,boxes,max_bindist)
 
 
@@ -125,7 +130,7 @@ def draw(self, regions, box ,hmargin,warnings=[]):
                             start2 = region2.start - right_offset *resolution
 
                 #start1,end1,left_offset = region1.start, region1.end,0
-                mat = c.matrix(balance=True).fetch(region1.chr+":"+str(start1)+"-"+str(end1),
+                mat = c.matrix(balance=self.balanced).fetch(region1.chr+":"+str(start1)+"-"+str(end1),
                                                 region2.chr+":"+str(start2)+"-"+str(end2))
                 mat[np.isnan(mat)]=0
                 if region1.orientation=="-": mat = mat[::-1,:]
@@ -222,7 +227,7 @@ def get_min_max_values(self,regions,low_percentile,high_percentile):
             for b in range(a,len(regions)):
                 if (not self.interactions_across_regions) and a!=b: continue
                 region1,region2 = correct_region_chr(regions[a],c.chromnames), correct_region_chr(regions[b],c.chromnames,file=self.file)
-                mat = c.matrix(balance=True).fetch(region1.chr+":"+str(region1.start)+"-"+str(region1.end),
+                mat = c.matrix(balance=self.balanced).fetch(region1.chr+":"+str(region1.start)+"-"+str(region1.end),
                                                 region2.chr+":"+str(region2.start)+"-"+str(region2.end))
                 mat[np.isnan(mat)]=0
                 if a!=b and self.double_interactions_across_regions: mat = 1+2*mat
@@ -237,9 +242,11 @@ def find_angle(self,regions,boxes,max_bindist):
         if boxes[0]["left"]>boxes[0]["right"]: min_angle=-100
         for a in range(len(regions)):
             region1 = correct_region_chr(regions[a],c.chromnames,file=self.file)
+            if not region1.chr in c.chromnames: raise KnownException("Could not find chromosome "+region1.chr+" in the .cool file. Please make sure that you specified the correct chromosome name (the chr prefix can be omitted). "\
+                                                                     "Only the following chromosome names were found in the .cool file: "+", ".join(c.chromnames))
             box1 = boxes[a]
             try:
-                mat = 1+c.matrix(balance=True).fetch(region1.chr+":"+str(region1.start)+"-"+str(region1.end))
+                mat = 1+c.matrix(balance=self.balanced).fetch(region1.chr+":"+str(region1.start)+"-"+str(region1.end))
             except ValueError:
                 raise KnownException("Could not retrieve region "+region1.chr+":"+str(region1.start)+"-"+str(region1.end)+" in file "+self.file+"."\
                                      " Make sure that the region that you specified does not extend beyond the chromosome length, and that you did not subset your .cool file.")

figeno/utils.py

@@ -14,6 +14,8 @@ def correct_region_chr(region,chromosomes,file=""):
         return region
     elif "chr"+region.chr in chromosomes:
         return Region("chr"+region.chr,region.start,region.end,region.orientation,region.color)
+    elif "Chr"+region.chr in chromosomes:
+        return Region("Chr"+region.chr,region.start,region.end,region.orientation,region.color)
     elif region.chr.lstrip("chr") in chromosomes:
         return Region(region.chr.lstrip("chr"),region.start,region.end,region.orientation,region.color)
     else: 

pyproject.toml

@@ -9,7 +9,7 @@ packages = ["figeno", "figeno.data", "figeno.cli", "figeno.gui"]
 
 [project]
 name = 'figeno'
-version = "1.3.2"
+version = "1.3.3"
 description = 'Package for generating genomics figures.'
 readme = 'README.md'
 authors = [