SlidingWin fix and new version

aphylogeo/alignement.py

@@ -183,8 +183,8 @@ def __init__(
         self.alignment_method = alignment_method
         self.reference_gene_file = reference_gene_file
         self.fit_method = fit_method
-        self.rate_similarity = (rate_similarity,)
-        self.method_similarity = (method_similarity,)
+        self.rate_similarity = rate_similarity
+        self.method_similarity = method_similarity
         """
         Method that align sequences
         """
@@ -726,7 +726,7 @@ def insertDash(self, dict, pos, keyList):
             dict[k] = s
         return dict
 
-    def slidingWindow(self, optimized=False):
+    def slidingWindow(self, optimized=True):
         """
         Method that slices all the sequences in a dictionary to a specific window (substring)
 
@@ -763,13 +763,13 @@ def slidingWindow(self, optimized=False):
                     windowed_alignment[f"{i}_{i + step - 1}"] = {key: val[i : i + step - 1] for key, val in paddedMSA.items()}
                     combinations = itertools.combinations(windowed_alignment[f"{i}_{i + step - 1}"].values(), 2)
                     df = pd.DataFrame(list(combinations))
-                    if self.rate_similarity[0] < self.similarity(df):
+                    if self.rate_similarity > self.similarity(df):
                         windowed_alignment.pop(f"{i}_{i + step - 1}")
                 else:
                     windowed_alignment[f"{i}_{seq_len-1}"] = {key: val[i : i + seq_len - 1] for key, val in paddedMSA.items()}
                     combinations = itertools.combinations(windowed_alignment[f"{i}_{seq_len-1}"].values(), 2)
                     df = pd.DataFrame(list(combinations))
-                    if self.rate_similarity[0] < self.similarity(df):
+                    if self.rate_similarity > self.similarity(df):
                         windowed_alignment.pop(f"{i}_{seq_len-1}")
         else:
             for i in range(0, seq_len, step):

aphylogeo/main.py

@@ -3,6 +3,7 @@
 
 from aphylogeo.params import Params
 from aphylogeo import utils
+from aphylogeo.genetic_trees import GeneticTrees
 
 # from aphylogeo.utils import climaticPipeline, geneticPipeline, filterResults, loadSequenceFile
 
@@ -24,11 +25,13 @@
     sequenceFile = utils.loadSequenceFile(Params().reference_gene_file)
     seq_alignment = AlignSequences(sequenceFile).align()
     # Phylo.write(tree1, "data/tree1.nwk", "newick")
-    # seq_alignment.save_to_json("./debug/sequences_aligned.json")
+    seq_alignment.save_to_json("./debug/sequences_aligned.json")
 
     # loaded_seq_alignment = Alignment.load_from_json("./debug/sequences_aligned.json")
 
     geneticTrees = utils.geneticPipeline(seq_alignment.msa)
+    trees = GeneticTrees(trees_dict=geneticTrees, format="newick")
+    trees.save_trees_to_json("./debug/geneticTreesTest.json")
     with open("debug/res3.txt", "w") as f:
         f.write(str(geneticTrees))
         f.close()

aphylogeo/params.yaml

@@ -14,7 +14,7 @@ alignment_method: '1' #Please select one ~ 1:pairwiseAligner, 2:MUSCLE, 3:CLUSTA
 distance_method: '1' #Please select one  ~ 1: Least-Square distance, 2: Robinson-Foulds distance, 3: Euclidean distance (DendroPY)
 fit_method: '1' #Please select one ~ 1:Wider Fit by elongating with Gap (starAlignment), 2:Narrow-fit prevent elongation with gap when possible
 tree_type: '1' #Please select one ~ 1: BioPython consensus tree, 2: FastTree application
-rate_similarity: 80
+rate_similarity: 90
 method_similarity: '1' #Please select one :
     # 1: Hamming distance
     # 2: Levenshtein distance

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "aphylogeo"
-version = "0.3.2"
+version = "0.4.2"
 description = "A phylogenetic and geographic analysis tool"
 authors = ["'Tahiri Lab'"]
 license = "MIT"