fixed rotations and inline ss

bayespairing/src/BayesPairing.py

@@ -480,13 +480,14 @@ def parse_sequence(seq, modules, ss, dataset, BNs, t=-10, samplesize=20000, Lamb
             #print(seq)
             #print(struct)
             if not str(r.node.positions) in modules_predicted[r.module.ID]:
+                #print("about to score",r.seq, r.rotation, r.pos)
                 r.eval(fold_compound=fc)
                 #print("SCORED MODULE",r.module.ID, r.seq, r.pos, r.score)
                 #print("PASSING THRESHOLD",r.score,t)
                 if r.score > t:
                     #print(r.module.ID, r.seq, r.pos, r.score)
                     #print("Adding module match to results")
-                    modules_predicted[r.module.ID][str(r.node.positions)] = [r.seq, r.pos,  prob,r.node.positions, r.score]
+                    modules_predicted[r.module.ID][str(r.node.positions)] = [r.seq, r.pos,  prob,r.node.positions, r.score, r.mapping]
                     
             else:
                 #print("ADDING PREDICTION TO OTHERS",r.pos,  modules_predicted[r.module.ID])

bayespairing/src/parse_sequences.py

@@ -743,7 +743,7 @@ if __name__ == "__main__":
 
             #print("THE HITS")
             #print(all_results[inputSeqKey])
-            modules_in_svg, chef_ss = bp_chefs_choice(all_results[inputSeqKey],seqInfo[seqCounter],arguments["t"],finalName)
+            modules_in_svg, chef_ss = bp_chefs_choice(all_results[inputSeqKey],seqInfo[seqCounter],float(arguments["t"]),finalName)
 
 
             #now we need to fill svg hits

bayespairing/src/pgmpy/models/RNAModule.py

@@ -325,7 +325,9 @@ class RNAModule(BayesianModel):
 
 
     def eval(self, sequence):
+        #print('testing seq against nodes',sequence,self.nodes())
         candidate = self.candidate_from_sequence(sequence)
+        #print('testing seq against nodes',sequence,self.nodes(),candidate)
         if candidate is None:
             return [-100, None]
         # in the bayesnet object, the nucleotide are stored as number
@@ -342,6 +344,7 @@ class RNAModule(BayesianModel):
             this_node_evidence_dict = {}
             for anc in ancestors_of_node:
                 this_node_evidence_dict[anc] = candidate[anc]
+            #print("evaluating",node, candidate[node], this_node_evidence_dict)
             node_probability = math.log(compute_probability(self, node, candidate[node], this_node_evidence_dict))
             node_probabilities.append(node_probability)
             module_probability = module_probability + node_probability
@@ -400,6 +403,22 @@ class RNAModule(BayesianModel):
         return candidate
 
 
+    def mapping_from_sequence(self, sequence):
+        #print("SCORING CANDIDATE", self.ID, self.nodes(),sequence, "GAPS",self.gaps)
+        #takes as input a sequence with stars, and outputs a dictionary of node values
+        candidate = {}
+        if "T" in sequence:
+            sequence.replace("T","U")
+
+        sequence = list(sequence)
+        sequence = [i for i in sequence if i.isalpha()]
+  
+        for ind,node in enumerate(self.nodes(data=False)):
+            candidate[node] = sequence[ind]
+        return candidate
+
+
+
     def eval_constraint_folding(self,seq_score,full_seq,positions,rotated_ss, rotated_gaps, fc):
         #print('testing candidate with constraint folding',rotated_ss, rotated_ss.split("*"))
 

bayespairing/src/scanning/src/classes/ScanResult.py

@@ -58,7 +58,7 @@ class ExactScanResult:
         if rotation:
             self.rotate(rotation)
         #print("LEN POS",self.pos,"LEN SEQ",self.seq, "ROTATION",rotation)
-
+        #print("DONE ROTATION, CURRENT SEQUENCE", self.seq)
 
         #TODO : need to think about adjusting POSITIONS to gaps
         self.gaps = self.module.gaps_per_strand
@@ -66,7 +66,8 @@ class ExactScanResult:
         
         #print("PREFIX SEQUENCES",self.seq)
         self.reintegrate_module_gaps_to_sse_sequence_and_positions()
-
+        #print("Reintegrated module graps, current seq", self.seq)
+        self.mapping = {}
 
 
     def rotate_stackings(self):
@@ -112,7 +113,8 @@ class ExactScanResult:
                     if ind not in gaps[0]:
                         new_seq += nuc
         else:
-            for strand,subseq in enumerate(self.rotate_list(seq.split("&"))):
+            #for strand,subseq in enumerate(self.rotate_list(seq.split("&"))):
+            for strand,subseq in enumerate(seq.split("&")):
                 #print("STRAND",strand,"GAPS",gaps, seq, seq.split("&"))
                 if not gaps[strand]:
                     new_seq+=subseq
@@ -147,7 +149,7 @@ class ExactScanResult:
         """
 
         #TODO: need to make sure the gaps are managed when scoring the sequence, specifically through rotation
-        #print("info",self.module.stackings,self.module.ID,self.module.n_nodes,self.seq)
+        #print("info eval",self.rotation,self.pos,self.module.ID,self.module.n_nodes,self.seq)
         try:
             return self.score
         except:
@@ -161,6 +163,7 @@ class ExactScanResult:
                     return -1
                # print(self.module.ID,self.node,self.pos)
                 score = self.module.eval(self.seq)[0]
+                self.mapping = self.module.mapping_from_sequence(self.seq)
                 #print("COMPUTING SEQUENCE SCORE",score, self.pos, self.seq)
 
                 self.score = score
@@ -182,6 +185,8 @@ class ExactScanResult:
                     else:
                         #print("Candidate scoring", tmp_seq,self.module.eval(tmp_seq)[0] )
                         total_score+= self.module.eval(tmp_seq)[0]
+                        if len(self.mapping)==0:
+                            self.mapping = self.module.mapping_from_sequence(tmp_seq)
                 mean_score = total_score/(len(sequences)-gapped_cand)
                 # print("score for this alignment:",mean_score)
                 self.score = mean_score