renamed aa_index folder to aa_index_scripts

scripts/aa_index_scripts/aaindex/data/parse_aaindex.py

@@ -0,0 +1,90 @@
+from collections import defaultdict
+
+import os
+import pickle
+
+DATA_FOLDER = "/home/chmrodrigues/Documents/ppi2/reverse_mutations/data/aaindex"
+
+def main():
+
+    aaindex2_file = os.path.join(DATA_FOLDER,"aaindex2")
+    aaindex3_file = os.path.join(DATA_FOLDER,"aaindex3")
+
+    lines_index2 = ' '.join([item for item in open(aaindex2_file,'r').readlines()])
+    lines_index3 = ' '.join([item for item in open(aaindex3_file,'r').readlines()])
+
+    attrs_index2 = [item for item in lines_index2.split('//\n') if len(item) != 0]
+    attrs_index3 = [item for item in lines_index3.split('//\n') if len(item) != 0]
+    
+    attr_name = str()
+    all_matrices = dict()
+    for line in attrs_index2:
+        attr_elements = line.split('\n')
+
+        attr_name = [item for item in attr_elements if item.strip().startswith("H ")][0].split()[-1]
+        rows_columns_index = [attr_elements.index(item) for item in attr_elements if item.startswith(" M rows =")][0]
+
+        rows = attr_elements[rows_columns_index].split()[3].replace(",","")
+        columns = attr_elements[rows_columns_index].split()[-1]
+
+        attr_dict = dict()
+        for row in rows:
+            attr_dict[row] = dict()
+            for col in columns:
+                attr_dict[row][col] = None
+
+        for i in range(rows_columns_index+1,len(attr_elements)):
+            values = attr_elements[i].split()
+            try:
+                row = rows[i-(rows_columns_index+1)]
+                for idx,value in enumerate(values):
+                    col = columns[idx]
+                    try:
+                        attr_dict[row][col] = float(value)
+                    except ValueError:
+                        attr_dict[row][col] = value
+            except IndexError:
+                pass
+		
+        all_matrices[attr_name] = attr_dict
+    print(len(all_matrices))
+    pickle.dump(all_matrices, open('index2.p','wb'),protocol=2)
+
+    attr_name = str()
+    all_matrices = dict()
+    for line in attrs_index3:
+        attr_elements = line.split('\n')
+
+        attr_name = [item for item in attr_elements if item.strip().startswith("H ")][0].split()[-1]
+        rows_columns_index = [attr_elements.index(item) for item in attr_elements if item.startswith(" M rows =")][0]
+
+        rows = attr_elements[rows_columns_index].split()[3].replace(",","")
+        columns = attr_elements[rows_columns_index].split()[-1]
+
+        attr_dict = dict()
+        for row in rows:
+            attr_dict[row] = dict()
+            for col in columns:
+                attr_dict[row][col] = None
+
+        for i in range(rows_columns_index+1,len(attr_elements)):
+            values = attr_elements[i].split()
+            try:
+                row = rows[i-(rows_columns_index+1)]
+                for idx,value in enumerate(values):
+                    col = columns[idx]
+                    try:
+                        attr_dict[row][col] = float(value)
+                    except ValueError:
+                        attr_dict[row][col] = value
+            except IndexError:
+                pass
+		
+        all_matrices[attr_name] = attr_dict
+    pickle.dump(all_matrices, open('index3.p','wb'),protocol=2)
+    print(len(all_matrices))
+
+    return True
+
+if __name__ == "__main__":
+    main()