implemented blast

detector.py

@@ -13,6 +13,20 @@
 RUN_BLAST_CMD = 'blastp -query %s -db %s -out %s -outfmt 6'
 
 
+def read_blast_line(line):
+    """
+    Parses a line of tabular blast output, returns a dict with the header
+
+    :param line: line from a blast file to parse
+    :return: dict with field names a keys and values the values from the line
+    """
+    header = ['qseqid', 'sseqid', 'pident', 'length', 'mismatch', 'gapopen', 'qstart', 'qend', 'sstart', 'send',
+              'evalue', 'bitscore']
+    types = [str, str, float, int, int, int, int, int, int, int, float, float]
+
+    return {h: t(v) for h, v, t in zip(header, line.strip().split(), types)}
+
+
 @click.group()
 def cli():
     """
@@ -49,7 +63,7 @@ def build(path, db):
 
     # find fasta files and concatenate them to a temporary file
     with open(full_fasta, 'w') as f_out:
-        for file in os.listdir(path):
+        for file in path:
             if file.lower().endswith('.fasta'):
                 click.echo("\tFound file: %s" % file)
                 fasta_count += 1
@@ -95,9 +109,28 @@ def blast(fasta, db, output):
 
 
 @cli.command()
-@click.argument('input', type=click.Path())
-def analyze(input):
+@click.argument('blast_results', type=click.File())
+@click.argument('output', type=click.File(mode='w'))
+def analyze(blast_results, output):
     """
     Analyzes blast output
     """
-    pass
+    click.secho("Analyzing file %s ... " % blast_results.name, fg='green', bold=True)
+
+    transposable_elements = set()
+
+    for line in blast_results:
+        data = read_blast_line(line)
+        if data['pident'] > 70:
+            transposable_elements.add(data['qseqid'])
+
+    if len(transposable_elements) > 0:
+        click.echo("Found %d potential TEs, writing list to %s ..." % (len(transposable_elements), output.name))
+
+    for te in transposable_elements:
+        print(te, file=output)
+
+    # done
+    click.secho("All done!", fg='green', bold=True)
+
+