supprimé fonctions non utilisées

Functions.py

@@ -273,40 +273,6 @@ def get_sequence_list_seg(list_segfeat_nb,i,feature,seg_seq):
             del_sequence+=seg_seq[list_segfeat_nb[k]]
     return del_sequence
 
-def get_old_new_pos_substitution(feat_start,list_segfeat_nb,list_segfeat_azu,feat,i,j):
-    pos_old=int(Segments[list_segfeat_nb[i]].start)-int(feat_start)+1
-
-    start_feat=get_feature_start_on_genome(list_segfeat_azu[0],feat)
-    start_var=int(segments_on_target_genome[list_segfeat_azu[j-1]][2])+1
-    pos_new=str(start_var-start_feat+1)
-    return [pos_old,pos_new]
-
-def get_old_new_pos_insertion(last_start,feat_start,list_segfeat_azu,feat):
-    pos_old=str(int(Segments[last_start].start)-int(feat_start)+1)
-
-    start_feat=get_feature_start_on_genome(list_segfeat_azu[0],feat) 
-    start_var=int(segments_on_target_genome[last_start][1])-1
-    pos_new=str(start_var-start_feat+1)
-    return [pos_old,pos_new]
-
-def get_old_new_pos_deletion(last_start,feat_start,list_segfeat_azu,feat,last_in_azu,i):
-    if i==0:
-        pos_old=int(Segments[last_start].start)-int(feat_start)+1+Features[feat].pos_start
-    else:
-        pos_old=int(Segments[last_start].start)-int(feat_start)+1
-        if pos_old<0:
-            pos_old=0
-
-    if last_in_azu=="": # deletion of the beggining of the feature, so no segment placed in the new genome yet. 
-        pos_new="1"
-    else:
-        start_feat=get_feature_start_on_genome(list_segfeat_azu[0],feat) 
-        start_var=int(segments_on_target_genome[last_in_azu][2])+1 
-        pos_new=str(start_var-start_feat+1)
-    
-    return [pos_old,pos_new]
-
-
 class Variation:
     def __init__(self,feature_id,feature_type,chr,start_new,stop_new,inversion,size_diff):
         self.feature_id=feature_id
@@ -349,7 +315,7 @@ def reset_var(variation):
     variation.ref=''
     variation.alt=''
 
-def get_old_new_pos_substitution_2(feat_start,variation,list_segfeat_azu,feat,j):
+def get_old_new_pos_substitution(feat_start,variation,list_segfeat_azu,feat,j):
     #pos_old=int(Segments[list_segfeat_nb[i]].start)-int(feat_start)+1
     pos_old=str(int(Segments[variation.start_var].start)-int(feat_start)+1)
     start_feat=get_feature_start_on_genome(list_segfeat_azu[0],feat)
@@ -357,7 +323,7 @@ def get_old_new_pos_substitution_2(feat_start,variation,list_segfeat_azu,feat,j)
     pos_new=str(start_var-start_feat+1)
     return [pos_old,pos_new]
 
-def get_old_new_pos_insertion_2(variation,feat_start,list_segfeat_azu,feat):
+def get_old_new_pos_insertion(variation,feat_start,list_segfeat_azu,feat):
     pos_old=str(int(Segments[variation.start_var].start)-int(feat_start)+1)
 
     start_feat=get_feature_start_on_genome(list_segfeat_azu[0],feat) 
@@ -365,7 +331,7 @@ def get_old_new_pos_insertion_2(variation,feat_start,list_segfeat_azu,feat):
     pos_new=str(start_var-start_feat+1)
     return [pos_old,pos_new]
 
-def get_old_new_pos_deletion_2(variation,feat_start,list_segfeat_azu,feat,i):
+def get_old_new_pos_deletion(variation,feat_start,list_segfeat_azu,feat,i):
     if i==0:
         pos_old=int(Segments[variation.start_var].start)-int(feat_start)+1+Features[feat].pos_start
     else:

Functions_output.py

@@ -112,7 +112,7 @@ def genome_gff(pos_seg, gff, out, gfa):
 
         # outputs the detail of variations of the feature :
         if var:
-            print_variations_2(first_seg,last_seg,feat,paths,seg_seq)
+            print_variations(first_seg,last_seg,feat,paths,seg_seq)
             write_line("",output_variations,True)
 
         if stats==True:
@@ -143,162 +143,6 @@ def genome_gff(pos_seg, gff, out, gfa):
 
 def print_variations(first_seg,last_seg,feat,paths,seg_seq):
 
-    if (first_seg!=''): # if the feature is not completly absent        # add the else, output absent features
-
-        var=0 # count variations, to see if there is any
-        feature=Features[feat]
-        feat_start=feature.start
-
-        # get the lengths of the feature, on the original genome and on the new one
-        start_new_genome=get_feature_start_on_genome(first_seg,feat)
-        stop_new_genome=get_feature_stop_on_genome(last_seg,feat)
-        size_new_genome=int(stop_new_genome)-int(start_new_genome)+1
-        size_diff=str(size_new_genome-feature.size)
-
-        # get feature paths on the original genome and on the target genome
-        list_segfeat_azu=get_feature_path(paths,first_seg,last_seg)
-        list_segfeat_nb=feature.segments_list
-
-        # loop to go through both paths
-        i=0
-        j=0
-        [last,last_taille,last_seq,last_start,last_in_azu]=['',0,'','',''] # rename last_taille 
-
-        # check if there is an inversion and remove strands
-        [list_segfeat_nb,list_segfeat_azu,inversion]=detect_inversion(list_segfeat_nb,list_segfeat_azu)
-
-        # detect and print variations ignoring the strands
-        while (i<len(list_segfeat_nb)) & (j<len(list_segfeat_azu)):
-            if list_segfeat_nb[i] != list_segfeat_azu[j]: # if there is a difference between the two paths
-                if list_segfeat_azu[j] not in list_segfeat_nb: # if the segment in azu is absent in nb
-                    if list_segfeat_nb[i] not in list_segfeat_azu: # if the segment in nb is absent is azu
-                        # is both segments are absent in the other genome, its a substitution
-                        last_in_azu=list_segfeat_azu[j]
-
-                        # print if we had an insertion or deletion running
-                        if last=='insertion': # last : type énuméré.    # fct compare cas précedent à cas courant, si différent imprimer cas précédent. 
-                            [pos_old,pos_new]=get_old_new_pos_insertion(last_start,feat_start,list_segfeat_azu,feat)
-                            # line : formated line f"{feat}\t"
-                            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tinsertion\t-\t"+last_seq+"\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                            write_line(line,output_variations,False)
-                            var+=1
-                        elif last=='deletion':
-                            [pos_old,pos_new]=get_old_new_pos_deletion(last_start,feat_start,list_segfeat_azu,feat,last_in_azu,i)
-
-                            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tdeletion\t"+last_seq+"\t-\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                            write_line(line,output_variations,False)                          
-                            var+=1
-                        last='';last_taille=0;last_seq='';last_start=''
-
-                        # print the substitution        # what if plusieurs substitutions à la suite ? 
-                        # substitution of segment list_segfeat_nb[i][1:] by segment list_segfeat_azu[j][1:]
-                        [pos_old,pos_new]=get_old_new_pos_substitution(feat_start,list_segfeat_nb,list_segfeat_azu,feat,i,j)
-
-                        if len(seg_seq[list_segfeat_nb[i]]) == len(seg_seq[list_segfeat_azu[j]]): # if the substituion is between two segment of the same size, print it
-                            size_subs=len(seg_seq[list_segfeat_nb[i]])
-                            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tsubstitution\t"+seg_seq[list_segfeat_nb[i]]+"\t"+seg_seq[list_segfeat_azu[j]]+"\t"+str(size_subs)+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                        else :
-                            # if the segments of the substitution have a different size, print deletion then insertion at the same position.
-                            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tdeletion\t"+seg_seq[list_segfeat_nb[i]]+"\t-\t"+str(len(seg_seq[list_segfeat_nb[i]]))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                            line+=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tinsertion\t-\t"+seg_seq[list_segfeat_azu[j]]+"\t"+str(len(seg_seq[list_segfeat_azu[j]]))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                            var+=1
-                        write_line(line,output_variations,False)
-                        var+=1;i+=1;j+=1
-
-                    else: # azu segment not in nb, but nb segment in azu : insertion
-                        if last=='deletion':
-                            [pos_old,pos_new]=get_old_new_pos_deletion(last_start,feat_start,list_segfeat_azu,feat,last_in_azu,i)
-
-                            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tdeletion\t"+last_seq+"\t-\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                            write_line(line,output_variations,False)
-                            var+=1;last='';last_taille=0;last_start='';last_seq=''
-
-                        last_in_azu=list_segfeat_azu[j]
-
-                        if last=='insertion':
-                            last_seq=last_seq+seg_seq[list_segfeat_azu[j]]
-                        else:
-                            last='insertion'
-                            last_seq=seg_seq[list_segfeat_azu[j]]
-                            last_start=list_segfeat_nb[i]
-                        j+=1
-                        
-                elif list_segfeat_nb[i] not in list_segfeat_azu: # nb segment not in azu, but azu segment in nb : deletion
-                    if last=='insertion':
-                        [pos_old,pos_new]=get_old_new_pos_insertion(last_start,feat_start,list_segfeat_azu,feat)
-                        
-                        line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tinsertion\t-\t"+last_seq+"\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                        write_line(line,output_variations,False)
-                        var+=1;last='';last_start='';last_taille=0;last_seq=''
-
-                    if last=='deletion':
-                        last_seq=last_seq+seg_seq[list_segfeat_nb[i]]
-                    else:
-                        last='deletion'
-                        last_start=list_segfeat_nb[i]
-                        if i==0: # if the deletion is at the start of the feature, the deletion doesnt start at the start at the first segment : 
-                                    #use pos_start, position of the feature on its first segment
-                            last_seq=seg_seq[list_segfeat_nb[i]][feature.pos_start-1:]
-                        else: # else, the deletion will always start at the start of the first segment.
-                            last_seq=seg_seq[list_segfeat_nb[i]]
-                    i+=1
-                    
-                else : # idk yet. if both segments are present in the other genome but not at the same position. probably substitution then
-                    line="weird order change\n"
-                    write_line(line,output_variations,False)
-                    var+=1;i+=1;j+=1
-
-            else: # segment present in both. print the running indel if there is one
-                if last=='insertion':
-                    [pos_old,pos_new]=get_old_new_pos_insertion(last_start,feat_start,list_segfeat_azu,feat)
-                    
-                    line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tinsertion\t-\t"+last_seq+"\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                    write_line(line,output_variations,False)
-                    var+=1
-
-                elif last=='deletion':
-                    [pos_old,pos_new]=get_old_new_pos_deletion(last_start,feat_start,list_segfeat_azu,feat,last_in_azu,i)
-                    
-                    line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tdeletion\t"+last_seq+"\t-\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-                    write_line(line,output_variations,False)
-                    var+=1
-
-                last_in_azu=list_segfeat_azu[j]
-                last='';last_taille=0;last_start='';last_seq='';i+=1;j+=1
-
-        # finish printing the current indel
-        if last=='insertion':
-            [pos_old,pos_new]=get_old_new_pos_insertion(last_start,feat_start,list_segfeat_azu,feat)
-
-            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tinsertion\t-\t"+last_seq+"\t"+str(last_taille)+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-            write_line(line,output_variations,False)
-            var+=1
-
-        elif last=='deletion':
-            [pos_old,pos_new]=get_old_new_pos_deletion(last_start,feat_start,list_segfeat_azu,feat,last_in_azu,i)
-
-            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tdeletion\t"+last_seq+"\t-\t"+str(len(last_seq))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-            write_line(line,output_variations,False)
-            var+=1
-
-        # see if the end is missing for one of the two genomes
-        if not((i>=len(list_segfeat_nb)-1) & (j>=len(list_segfeat_azu)-1)):
-            pos_old=int(Segments[list_segfeat_nb[i]].start)-int(feat_start)+1
-            del_sequence=get_sequence_list_seg(list_segfeat_nb,i,feature,seg_seq)
-            length=len(del_sequence)
-            pos_new=str(size_new_genome+1) # the deletion is at the end of the feature on the new genome
-
-            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tdeletion\t"+del_sequence+"\t-\t"+str(length)+"\t"+str(pos_old)+"\t"+pos_new+"\n"
-            write_line(line,output_variations,False)
-            var+=1
-
-        if var==0:
-            line=feat+"\t"+feature.type+"\t"+feature.chr+"\t"+str(start_new_genome)+"\t"+str(stop_new_genome)+"\t"+str(size_new_genome)+"\t"+str(inversion)+"\t"+size_diff+"\tno_var\t-\t-\t-\t-\t-\n"
-            write_line(line,output_variations,False)
-
-
-def print_variations_2(first_seg,last_seg,feat,paths,seg_seq):
-
     if (first_seg!=''): # if the feature is not completly absent        # add the else, output absent features
         [variation,list_segfeat_nb,list_segfeat_azu]=create_var(feat,first_seg,last_seg,paths) # removes the strands in the segment lists
         feature=Features[feat]
@@ -376,15 +220,15 @@ def print_variations_2(first_seg,last_seg,feat,paths,seg_seq):
 
 def print_current_var(variation,feat_start,list_segfeat_azu,feat,i,j):
     if variation.type=='insertion':
-        [pos_old,pos_new]=get_old_new_pos_insertion_2(variation,feat_start,list_segfeat_azu,feat)
+        [pos_old,pos_new]=get_old_new_pos_insertion(variation,feat_start,list_segfeat_azu,feat)
         line=variation.feature_id+"\t"+variation.feature_type+"\t"+variation.chr+"\t"+str(variation.start_new)+"\t"+str(variation.stop_new)+"\t"+variation.size_new+"\t"+variation.inversion+"\t"+variation.size_diff+"\tinsertion\t-\t"+variation.alt+"\t"+str(len(variation.alt))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
         write_line(line,output_variations,False)
     elif variation.type=='deletion':
-        [pos_old,pos_new]=get_old_new_pos_deletion_2(variation,feat_start,list_segfeat_azu,feat,i)
+        [pos_old,pos_new]=get_old_new_pos_deletion(variation,feat_start,list_segfeat_azu,feat,i)
         line=variation.feature_id+"\t"+variation.feature_type+"\t"+variation.chr+"\t"+str(variation.start_new)+"\t"+str(variation.stop_new)+"\t"+variation.size_new+"\t"+variation.inversion+"\t"+variation.size_diff+"\tdeletion\t"+variation.ref+"\t-\t"+str(len(variation.ref))+"\t"+str(pos_old)+"\t"+pos_new+"\n"
         write_line(line,output_variations,False)
     elif variation.type=='substitution':
-        [pos_old,pos_new]=get_old_new_pos_substitution_2(feat_start,variation,list_segfeat_azu,feat,j)
+        [pos_old,pos_new]=get_old_new_pos_substitution(feat_start,variation,list_segfeat_azu,feat,j)
         if len(variation.ref) == len(variation.alt): # if the substituion is between two segment of the same size, print it
             size_subs=len(variation.ref)
             line=variation.feature_id+"\t"+variation.feature_type+"\t"+variation.chr+"\t"+str(variation.start_new)+"\t"+str(variation.stop_new)+"\t"+variation.size_new+"\t"+variation.inversion+"\t"+variation.size_diff+"\tsubstitution\t"+variation.ref+"\t"+variation.alt+"\t"+str(size_subs)+"\t"+str(pos_old)+"\t"+pos_new+"\n"