ssi-analysis-result-parsers 0.0.4__tar.gz → 0.0.5__tar.gz

{ssi_analysis_result_parsers-0.0.4/ssi_analysis_result_parsers.egg-info → ssi_analysis_result_parsers-0.0.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ssi_analysis_result_parsers
-Version: 0.0.4
+Version: 0.0.5
 Summary: TODO
 Home-page: https://github.com/thej-ssi/ssi_analysis_result_parsers
 Author: Thor Bech Johannesen

{ssi_analysis_result_parsers-0.0.4 → ssi_analysis_result_parsers-0.0.5}/settings.ini

@@ -5,7 +5,7 @@
 ### Python library ###
 repo = ssi_analysis_result_parsers
 lib_name = %(repo)s
-version = 0.0.4
+version = 0.0.5
 min_python = 3.9
 license = MIT
 black_formatting = True

{ssi_analysis_result_parsers-0.0.4 → ssi_analysis_result_parsers-0.0.5}/ssi_analysis_result_parsers/Legionella_parser.py

@@ -25,8 +25,10 @@ import json  # for nicely printing json and yaml
 # import functions from core module (optional, but most likely needed).
 from ssi_analysis_result_parsers import (
     core,
+    blast_parser,
 )
-from .blast_parser import extract_presence_absence
+
+# from ssi_analysis_result_parsers.blast_parser import extract_presence_absence
 
 # Project specific libraries
 from pathlib import Path
@@ -39,11 +41,18 @@ def extract_legionella_sbt(legionella_sbt_results_tsv: Path) -> dict:
     Returns dictionary of results found in the Legionella SBT summary output
     """
     if os.path.exists(legionella_sbt_results_tsv):
-        df = pandas.read_csv(legionella_sbt_results_tsv, sep="\t")
-        df.set_index("sample", inplace=True, drop=True)
-        d = df.to_dict(orient="index")
-        fname = next(iter(d))
-        return d[fname]
+        try:
+            df = pandas.read_csv(legionella_sbt_results_tsv, sep="\t")
+            df.set_index("sample", inplace=True, drop=True)
+            d = df.to_dict(orient="index")
+            fname = next(iter(d))
+            return d[fname]
+        except pandas.errors.EmptyDataError:
+            print(
+                f"No Legionella SBT output empty at {legionella_sbt_results_tsv}",
+                file=sys.stderr,
+            )
+            return None
     else:
         print(
             f"No Legionella SBT output found at {legionella_sbt_results_tsv}",
@@ -56,7 +65,7 @@ def legionella_summary(legionella_sbt_results_tsv: Path, lag1_blast_tsv: Path) -
     sbt_results_dict = extract_legionella_sbt(
         legionella_sbt_results_tsv=legionella_sbt_results_tsv
     )
-    lag1_blast_dict = extract_presence_absence(
+    lag1_blast_dict = blast_parser.extract_presence_absence(
         blast_output_tsv=lag1_blast_tsv,
         hits_as_string=False,
         include_match_stats=False,
@@ -135,13 +144,14 @@ class LegionellaResults(core.PipelineResults):
             results_dict[sample_name] = legionella_results
         return cls(results_dict)"""
 
+    @staticmethod
     def legionella_summary(
         legionella_sbt_results_tsv: Path, lag1_blast_tsv: Path
     ) -> dict:
         sbt_results_dict = extract_legionella_sbt(
             legionella_sbt_results_tsv=legionella_sbt_results_tsv
         )
-        lag1_blast_dict = extract_presence_absence(
+        lag1_blast_dict = blast_parser.extract_presence_absence(
             blast_output_tsv=lag1_blast_tsv,
             hits_as_string=False,
             include_match_stats=False,
@@ -150,6 +160,8 @@ class LegionellaResults(core.PipelineResults):
         results_dict = core.update_results_dict(
             sbt_results_dict, lag1_blast_dict, old_duplicate_key_prefix="SBT: "
         )
+        if results_dict is None:
+            return {}
         return results_dict
 
     def __repr__(self):

ssi_analysis_result_parsers-0.0.5/ssi_analysis_result_parsers/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.0.5"

{ssi_analysis_result_parsers-0.0.4 → ssi_analysis_result_parsers-0.0.5}/ssi_analysis_result_parsers/blast_parser.py

@@ -51,21 +51,31 @@ def extract_presence_absence(
     if os.path.exists(blast_output_tsv):
         try:
             blast_df = pandas.read_csv(blast_output_tsv, sep="\t", header=None)
+            header_list = tsv_header.split(" ")
+            if len(header_list) == len(blast_df.columns):
+                blast_df.columns = tsv_header.split(" ")
+                blast_df["plen"] = blast_df["length"] / blast_df["qlen"] * 100
+                blast_df_unique = (
+                    blast_df.sort_values(by=["bitscore"], ascending=False)
+                    .groupby("qseqid")
+                    .first()
+                )
+                blast_df_filtered = blast_df_unique.query(
+                    "plen > @plen_threshold and pident > @pident_threshold"
+                )
+                blast_dict = dict(blast_df_filtered.to_dict(orient="index"))
+            else:
+                print(
+                    f"Failed to parse {blast_output_tsv}. Number of columns do not match length of provided header string",
+                    file=sys.stderr,
+                )
+                return None
 
-            blast_df.columns = tsv_header.split(" ")
-            blast_df["plen"] = blast_df["length"] / blast_df["qlen"] * 100
-            blast_df_unique = (
-                blast_df.sort_values(by=["bitscore"], ascending=False)
-                .groupby("qseqid")
-                .first()
-            )
-            blast_df_filtered = blast_df_unique.query(
-                "plen > @plen_threshold and pident > @pident_threshold"
-            )
-            blast_dict = dict(blast_df_filtered.to_dict(orient="index"))
         except pandas.errors.EmptyDataError:
             blast_dict = {}
             print(f"Blast output file {blast_output_tsv} empty. Assuming 0 blast hits.")
+        except Exception as e:
+            print(f"Error parsing blast: e")
         if hits_as_string:
 
             results = []

{ssi_analysis_result_parsers-0.0.4 → ssi_analysis_result_parsers-0.0.5/ssi_analysis_result_parsers.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ssi_analysis_result_parsers
-Version: 0.0.4
+Version: 0.0.5
 Summary: TODO
 Home-page: https://github.com/thej-ssi/ssi_analysis_result_parsers
 Author: Thor Bech Johannesen

{ssi_analysis_result_parsers-0.0.4 → ssi_analysis_result_parsers-0.0.5}/ssi_analysis_result_parsers.egg-info/SOURCES.txt

@@ -21,6 +21,7 @@ ssi_analysis_result_parsers.egg-info/top_level.txt
 ssi_analysis_result_parsers/config/config.default.env
 ssi_analysis_result_parsers/config/config.default.yaml
 test_input/.DS_Store
+test_input/empty_file.txt
 test_input/Legionella/batch_parser_file_paths.tsv
 test_input/Legionella/lag-1_blast.tsv
 test_input/Legionella/lag-1_blast_2.tsv

{ssi_analysis_result_parsers-0.0.4 → ssi_analysis_result_parsers-0.0.5}/test_input/Legionella/batch_parser_file_paths.tsv

@@ -1,3 +1,5 @@
 sample_name	sbt_results	lag1_blast_results
 sample_1	test_input/Legionella/test.sbt.tsv	test_input/Legionella/lag-1_blast.tsv
 sample_2	test_input/Legionella/test2.sbt.tsv	test_input/Legionella/lag-1_blast_2.tsv
+sample_3	test_input/Legionella/test2.sbt.tsv	test_input/empty_file.txt
+sample_4	test_input/empty_file.txt	test_input/Legionella/lag-1_blast_2.tsv

ssi_analysis_result_parsers-0.0.4/ssi_analysis_result_parsers/__init__.py

@@ -1 +0,0 @@
-__version__ = "0.0.3"