unaligned queries script and report

Author: tfursten

mtsv/commands/snakefiles/binning_snek

@@ -7,7 +7,9 @@ from mtsv.parsing import (
     format_cml_params,
     file_type, outfile_type,
     make_table)
-from mtsv.utils import bin_path, track_file_params, get_database_params
+from mtsv.utils import (
+    bin_path, script_path,
+    track_file_params, get_database_params)
 
 
 shell.prefix("set -euo pipefail;")
@@ -60,8 +62,10 @@ try:
 except FileNotFoundError:
     SAMPLES = config['fastq']
 SAMPLES = [os.path.basename(s) for s in SAMPLES]
+SAMPLE_NAMES = [os.path.splitext(s)[0] for s in SAMPLES]
 UNALIGNED_QUERIES = [os.path.join(config['binning_outpath'],
-            "unaligned_queries_{}.fasta".format(s)) for s in SAMPLES]
+            "unaligned_queries_{}.fasta".format(s)) for s in SAMPLE_NAMES]
+
 
 rule binning_all:
     input: REPORT
@@ -73,7 +77,8 @@ rule binning_report:
     output:
         rep=REPORT
     params:
-        header = SAMPLES
+        header = SAMPLE_NAMES,
+        merge_file = config['merge_file']
     message: 
         """
         Running binning report.
@@ -102,21 +107,29 @@ rule binning_report:
             "Unique Unaligned Queries"])
 
         file_str = "\n".join(UNALIGNED_QUERIES)
+        total_hits = summary_data['total_hits']
+        total_unique_hits = summary_data['total_unique_hits']
+        total_queries = summary_data['total_queries']
+        total_unique_queries = summary_data['total_unique_queries']
+        total_unaligned_queries = summary_data['total_unaligned_queries']
+        total_unique_unaligned_queries = summary_data[
+            'total_unique_unaligned_queries']
+
         track_file_params('merge_file', config['merge_file'], config)
 
         report("""
         Binning Report
         ================================
         Hits are reported in:\n
-        {config['merge_file']}\n
-        There were **{summary_data['total_hits']}** total hits
-        and **{summary_data['total_unique_hits']}** unique hits out 
-        of **{summary_data['total_queries']}** total queries and
-        **{summary_data['total_unique_queries']}** unique queries.\n
-        There were **{summary_data['total_unaligned_queries']}** total
-        queries and **{summary_data['total_unique_unaligned_queries']}**
+        **{params.merge_file}**\n
+        There were **{total_hits}** total hits
+        and **{total_unique_hits}** unique hits out 
+        of **{total_queries}** total queries and
+        **{total_unique_queries}** unique queries.\n
+        There were **{total_unaligned_queries}** total
+        queries and **{total_unique_unaligned_queries}**
         unique queries with no hits.\n
-        Analigned query sequences are reported in:\n
+        Unaligned query sequences are reported in:\n
         {file_str}.\n
         Query Stats:\n
         {table}
@@ -132,13 +145,15 @@ rule unaligned_queries:
         clps = config['merge_file'],
         fasta = config['fasta']
     params:
-        samples = SAMPLES
+        samples = SAMPLE_NAMES,
+        outpath = config['binning_outpath']
     threads: config['threads']
     log: config['log_file']
     message:
         """
         Finding unaligned queries.
-        Writing unaligned queries to {output}. 
+        Writing unaligned queries to {output}.
+        Logging to {log}.
         Snakemake scheduler assuming {threads} thread(s)
         """
     script: SCRIPT['unaligned_queries']

mtsv/scripts/MTSv_unaligned_queries.py

@@ -7,23 +7,10 @@
 from itertools import chain
 from multiprocessing import Pool
 from Bio import SeqIO
-from mtsv.utils import config_logging, error, line_generator#, fasta_generator
+from mtsv.utils import config_logging, error, line_generator, fasta_generator
 from mtsv.parsing import file_type, outpath_type, parse_query_id, outfile_type
 
 
-def fasta_generator(file_name, n_records):
-    with open(file_name, 'r') as handle:
-        records = {}
-        for record in SeqIO.parse(handle, 'fasta'):
-            records[record.id] = record
-            if len(records) == n_records:
-                yield records
-                records = {}
-        if len(records):
-            yield records
-        return
-
-
 class Result:
     def __init__(self, all_queries, unaligned_queries):
         self._all_queries = all_queries
@@ -288,7 +275,7 @@ def parse_args():
         config_logging(snakemake.log[0], "INFO")      
         LOGGER = logging.getLogger(__name__)
         get_unaligned_queries(snakemake.input[0],
-            snakemake.input[1], snakemake.output[0],
+            snakemake.input[1], snakemake.params[1],
             snakemake.params[0], snakemake.threads)
     except NameError:
         ARGS = parse_args()