generator

.gitignore

 *.pyc
 python/*.pyc
 data
+tmp
 

nested/core/generator.py

+#!/usr/bin/env python3
+
+import random
+
+from Bio import SeqIO
+from Bio.SeqRecord import SeqRecord
+
+from nested.core.te import TE
+from nested.core.nested_element import NestedElement
+from nested.utils import intervals
+
+class Generator(object):
+    """Class used to generate artificial nested elements for testing purposes
+
+    Attributes:
+        sourceFile (str): path to source database of TE's in fasta format
+    """
+
+    def __init__(self, source_db):
+        self._source_db = source_db
+        self.elements = []
+
+    def generate_random_nested_element(self, baselength=0, number_of_iterations=1, filter_string='LTR'):
+        sequences = list(SeqIO.parse(open(self._source_db), 'fasta'))
+        sequences = list(filter(lambda x: filter_string in x.id, sequences))
+
+        #random initial sequence
+        element_sequence = ''.join([random.choice('atgc') for x in range(baselength)])
+        nested_intervals = [[0, len(element_sequence) - 1]]
+
+        for i in range(number_of_iterations):
+            chosen_id = random.randint(0, len(sequences) - 1)
+            chosen_seq = sequences[chosen_id].seq
+            #print('Chosen sequence ({}) with length {}'.format(sequences[chosen_id].id, len(chosen_seq)))
+            insert_position = random.randint(0, len(element_sequence) - 1)
+            nested_intervals = [[insert_position, insert_position + len(chosen_seq) - 1]] + nested_intervals
+            element_sequence = element_sequence[:insert_position] + chosen_seq + element_sequence[insert_position:]
+
+        #expand intervals
+        nested_intervals = intervals.expand_list(nested_intervals)
+        nested_tes = []
+        for interval in nested_intervals:
+            nested_tes.append(TE(
+                location=interval
+            ))
+        
+        element_id = 'GENERATED_{}'.format(len(self.elements) + 1)
+        element = NestedElement(element_id, element_sequence, nested_tes)
+        self.elements.append(element)       
+
+    def generate_random_nested_elements(self, number_of_elements=1, baselength=0, number_of_iterations=1, filter_string='LTR'):
+        for i in range(number_of_elements):
+            self.generate_random_nested_element(baselength=baselength, number_of_iterations=number_of_iterations, filter_string=filter_string)
+
+    def save_elements_to_fasta(self, filepath):
+        sequences = [SeqRecord(x.sequence, id=x.id, description='') for x in self.elements]
+        with open(filepath, 'w') as output_handle:
+            SeqIO.write(sequences, output_handle, 'fasta')
+        
\ No newline at end of file

nested/core/te.py

@@ -20,7 +20,7 @@ class TE(object):
         features (dict): dictionary of features assigned to TE (i.e. list of domains on the optimal path in graph)
         score (float): evaluated score of TE
     """
-    def __init__(self, ppt=None, pbs=None, location=None, ltr_left_location=None, ltr_right_location=None, features=None, score=None):
+    def __init__(self, ppt=None, pbs=None, location=None, ltr_left_location=None, ltr_right_location=None, features={}, score=None):
         self.ppt = ppt
         self.pbs = pbs
         self.location = location

nested/utils/intervals.py

@@ -77,13 +77,13 @@ def expand(source, target):
         target[1] += length
     return target
 
-def expandList(intervalList):
+def expand_list(interval_list):
     """Expand intervals by the prevoius ones starting from the end
 
     Arguments:
-        intervalList(list): list of intervals to expand
+        interval_list(list): list of intervals to expand
     """
-    for i in reversed(range(len(intervalList) - 1)):
-        for j in range(i + 1, len(intervalList)):
-            intervalList[j] = expand(intervalList[i], intervalList[j])
-    return intervalList
\ No newline at end of file
+    for i in reversed(range(len(interval_list) - 1)):
+        for j in range(i + 1, len(interval_list)):
+            interval_list[j] = expand(interval_list[i], interval_list[j])
+    return interval_list
\ No newline at end of file