Add avg, improve fasta scripts

Author: enormandeau

Diff for: avg

@@ -0,0 +1,3 @@
+#!/bin/bash
+# Return the total of a file or stream with one number per line
+awk '{t+=$1} END {print t/NR}'

Diff for: fasta_extract_min_length.py

@@ -49,11 +49,11 @@ def fasta_iterator(input_file):
                     yield Fasta(name, "".join(sequence))
 
                 name = line[1:]
-                sequence = ""
+                sequence = []
                 begun = True
 
             else:
-                sequence += line
+                sequence.append(line)
 
         if name != "":
             yield Fasta(name, "".join(sequence))
@@ -69,8 +69,7 @@ def fasta_iterator(input_file):
 
 fasta_sequences = fasta_iterator(input_file)
 
-with open(output_file, "wt") as outfile:
+with myopen(output_file, "wt") as outfile:
     for seq in fasta_sequences:
         if len(seq.sequence) >= min_length:
             seq.write_to_file(outfile)
-

Diff for: fasta_n50.py

@@ -30,26 +30,30 @@ def myopen(_file, mode="rt"):
     else:
         return open(_file, mode=mode)
 
-def fasta_iterator(genome_file):
-    """Takes a fasta file genome_file and returns a fasta iterator
+def fasta_iterator(input_file):
+    """Takes a fasta file input_file and returns a fasta iterator
     """
-    with myopen(genome_file) as f:
-        sequence = ""
+    with myopen(input_file) as f:
+        sequence = []
         name = ""
         begun = False
+
         for line in f:
             line = line.strip()
+
             if line.startswith(">"):
                 if begun:
-                    yield Fasta(name, sequence)
-                name = line.replace(">", "")
+                    yield Fasta(name, "".join(sequence))
+
+                name = line[1:]
                 sequence = ""
                 begun = True
+
             else:
                 sequence += line
 
         if name != "":
-            yield Fasta(name, sequence)
+            yield Fasta(name, "".join(sequence))
 
 if __name__ == '__main__':
     # Prevent broken pipe error