fixed remnants of chdir bugs

bayespairing/src/make_BN_from_carnaval.py

@@ -131,7 +131,7 @@ def call_makeBN(mod,dataset,left_out):
         removecall = "rm "+ +"../models/"+bayesname
         subprocess.call(removecall, shell=True)
     #if not os.path.isfile(bayesname), changed for cross-validation; always remake the model
-    if not os.path.isfile(bayesname):
+    if not os.path.isfile("../models/"+bayesname):
          aln_list = pickle.load(open("../models/"+dataset  + "_aligned_modulegraphs.cPickle",'rb'))
          g_list = pickle.load(open("../models/"+dataset + "_one_of_each_graph.cPickle",'rb'))
          pdb_names  = pickle.load(open("../models/"+dataset + "_PDB_names.cPickle",'rb'))
@@ -163,7 +163,8 @@ def call_makeBN(mod,dataset,left_out):
          motif = make_graph_from_carnaval(g, aln_dict)
          pwm = BN_builder.build_pwm(motif.nodes(),aln_dict)
          bob  = BN_builder.build_BN(motif,pwm,aln_dict)
-         pickle.dump(bob,open(bayesname,'wb'))
+         #print("STORING BAYES NET")
+         pickle.dump(bob,open("../models/"+bayesname,'wb'))
          #pdb_list = get_pdb_list(pdb_names[current_ID])
          #pickle.dump(seqs,open("seqs_"+str(bayesname),'wb'))
 

bayespairing/src/parse_sequences.py

@@ -233,7 +233,7 @@ def run_fasta(input, modules_to_parse, dataset, ss="", arguments={}):
         if "T" in input:
             input = input.upper()
             input = str(input).replace("T", "U")
-            sequences = [input]
+        sequences = [input]
         if len(input) <= 200:
             maxs = run_BP(input, ss, modules_to_parse, dataset, "NONE", m, n, sm, mc, p, k,sscons)
             all_maxes = []
@@ -412,7 +412,7 @@ if __name__ == "__main__":
         modules_to_check = range(len(graphs))
 
     # executes BayesPairing on the sequence
-    print(run_fasta(seq, modules_to_check, dataset, ss, arguments))
+    print(run_fasta(seq, modules_to_check, dataset, ss, arguments)[0])
 
 #if __name__ != "__main__":
 #    parse_sequences("AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA")

bayespairing/test/tpp_riboswitch.ipynb

@@ -135,11 +135,11 @@
 %% Cell type:code id: tags:
 ``` python
 %run ../src/parse_sequences.py -d rna3dmotif \
--seq "GGACUCGGGGUGCCCUUCUGCGUGAAGGCUGAGAAAUACCCGUAUCACCUGAUCUGGAUAAUGCCAGCGUAGGGAAGUUC" -t 14 -p 25000 -o normal_modules
+-seq "GGACUCGGGGUGCCCUUCUGCGUGAAGGCUGAGAAAUACCCGUAUCACCUGAUCUGGAUAAUGCCAGCGUAGGGAAGUUC" -t 15.7 -p 25000 -o normal_modules
 ```
 %% Cell type:markdown id: tags:
 ## Assessing false positive predictions
@@ -515,10 +515,11 @@
 3D2V junction (thaliana form : 68/72
 T - loop : 108/109
 We can observe that low sample size leads to imprecise results, but as we get closer to 100,000 sample size, the results of the cross-validation are very close to what was reported so far in this notebook, which makes sense since the signal is mostly encoded in the relative position of the nucleotides, more than it is in their exact sequence.
+In total, 81 of the 109 sequences show signal for the combination of the three-way junction module and the T-loop with cross-validation.
 %% Cell type:code id: tags:
 ``` python
 import pickle