fix: force parsing specific columns as float

taxprofiler · Jul 13, 2023 · f252fb1 · f252fb1
1 parent 6407ec1
commit f252fb1
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 1 deletion.
diff --git a/src/taxpasta/infrastructure/application/krakenuniq/krakenuniq_profile_reader.py b/src/taxpasta/infrastructure/application/krakenuniq/krakenuniq_profile_reader.py
@@ -52,7 +52,11 @@ def read(cls, profile: BufferOrFilepath) -> DataFrame[KrakenUniqProfile]:
             header=0,
             index_col=False,
             skipinitialspace=True,
-            dtype={"%": float},
+            dtype={
+                KrakenUniqProfile.percent: float,
+                KrakenUniqProfile.duplicates: float,
+                KrakenUniqProfile.coverage: float,
+            },
         )
         cls._check_num_columns(result, KrakenUniqProfile)
         return result
diff --git a/tests/data/krakenuniq/ERR3201952.krakenuniq.report.txt b/tests/data/krakenuniq/ERR3201952.krakenuniq.report.txt
@@ -0,0 +1,4 @@
+# KrakenUniq v1.0.2 DATE:2023-07-13T11:10:57Z DB:./testdb-krakenuniq DB_SIZE:358660 WD:/home/james/git/nf-core/taxprofiler/testing/blah/work/d4/9641b418650bf507e0b05143a5800d
+# CL:/usr/local/bin/krakenuniq --db testdb-krakenuniq --threads 2 --report-file ERR3201952.krakenuniq.report.txt ERR3201952.merged.fastq.gz
+%	reads	taxReads	kmers	dup	cov	taxID	rank	taxName
+100	4937	4937	10648888	1	NA	0	no rank	unclassified
diff --git a/tests/integration/test_krakenuniq_etl.py b/tests/integration/test_krakenuniq_etl.py
@@ -55,6 +55,7 @@ def other_profile(data_dir: Path, request: pytest.FixtureRequest) -> Path:
     [
         "test3.krakenuniq.report.txt",
         "test1.krakenuniq.report.txt",
+        "ERR3201952.krakenuniq.report.txt",
         pytest.param(
             "test1-invalid.krakenuniq.report.txt",
             marks=pytest.mark.raises(exception=SchemaErrors),