Adds --aligner-args option for passing arguments to the aligner

src/hostile/aligner.py

@@ -68,6 +68,7 @@ def gen_clean_cmd(
         index: Path | None,
         rename: bool,
         sort_by_name: bool,
+        aligner_args: str,
         threads: int,
         force: bool,
     ) -> str:
@@ -90,6 +91,7 @@ def gen_clean_cmd(
             "{REF_ARCHIVE_PATH}": str(self.ref_archive_path),
             "{INDEX_PATH}": str(self.idx_path),
             "{FASTQ}": str(fastq),
+            "{ALIGNER_ARGS}": str(aligner_args),
             "{THREADS}": str(threads),
         }
         alignment_cmd = self.cmd
@@ -99,7 +101,7 @@ def gen_clean_cmd(
         rename_cmd = (
             # ' | awk \'BEGIN{{FS=OFS="\\t"}} {{$1=int(NR)" "; print $0}}\''
             # Skips header lines (starting with @) and begins counter from first record
-            '| awk \'BEGIN {{ FS=OFS="\\t"; line_count=0 }} /^@/ {{ next }} {{ $1=int(line_count+1)" "; print $0; line_count++ }}\''
+            ' | awk \'BEGIN {{ FS=OFS="\\t"; line_count=0 }} /^@/ {{ next }} {{ $1=int(line_count+1)" "; print $0; line_count++ }}\''
             if rename
             else ""
         )
@@ -129,6 +131,7 @@ def gen_paired_clean_cmd(
         index: Path | None,
         rename: bool,
         sort_by_name: bool,
+        aligner_args: str,
         threads: int,
         force: bool,
     ) -> str:
@@ -154,14 +157,15 @@ def gen_paired_clean_cmd(
             "{INDEX_PATH}": str(self.idx_path),
             "{FASTQ1}": str(fastq1),
             "{FASTQ2}": str(fastq2),
+            "{ALIGNER_ARGS}": str(aligner_args),
             "{THREADS}": str(threads),
         }
         alignment_cmd = self.paired_cmd
         for k in cmd_template.keys():
             alignment_cmd = alignment_cmd.replace(k, cmd_template[k])
         sort_cmd = " | samtools sort -n -O sam -@ 6 -m 1G" if sort_by_name else ""
         rename_cmd = (
-            # f' | awk \'BEGIN{{FS=OFS="\\t"; start=0}} /^@/{{next}} !start && !/^@/{{start=1}} start{{$1=int((NR+1)/2)" "; print $0}}\''
+            # ' | awk \'BEGIN{{FS=OFS="\\t"; start=0}} /^@/{{next}} !start && !/^@/{{start=1}} start{{$1=int((NR+1)/2)" "; print $0}}\''
             # Skips header lines (starting with @) and begins counter from first record
             ' | awk \'BEGIN {{ FS=OFS="\\t"; start=0; line_count=1 }} /^@/ {{ next }} !start && !/^@/ {{ start=1 }} start {{ $1=int((line_count+1)/2)" "; print $0; line_count++ }}\''
             if rename

src/hostile/cli.py

@@ -26,6 +26,7 @@ def clean(
     sort_by_name: bool = False,
     out_dir: Path = lib.CWD,
     threads: int = lib.THREADS,
+    aligner_args: str = "",
     force: bool = False,
     debug: bool = False,
 ) -> None:
@@ -40,6 +41,7 @@ def clean(
     :arg sort_by_name: sort reads by name
     :arg out_dir: path to output directory
     :arg threads: number of threads to use
+    :arg aligner_args: additional arguments for alignment
     :arg force: overwrite existing output files
     :arg debug: show debug messages
     """
@@ -64,6 +66,7 @@ def clean(
             sort_by_name=sort_by_name,
             out_dir=out_dir,
             aligner=aligner_paired,
+            aligner_args=aligner_args,
             threads=threads,
             force=force,
         )
@@ -75,6 +78,7 @@ def clean(
             sort_by_name=sort_by_name,
             out_dir=out_dir,
             aligner=aligner_unpaired,
+            aligner_args=aligner_args,
             threads=threads,
             force=force,
         )

src/hostile/lib.py

@@ -35,10 +35,13 @@
             # cdn_base_url="http://localhost:8000",  # python -m http.server
             cdn_base_url=f"https://objectstorage.uk-london-1.oraclecloud.com/n/lrbvkel2wjot/b/human-genome-bucket/o",
             data_dir=XDG_DATA_DIR,
-            cmd=("{BIN_PATH} -x '{INDEX_PATH}' -U '{FASTQ}'" " -k 1 --mm -p {THREADS}"),
+            cmd=(
+                "{BIN_PATH} -x '{INDEX_PATH}' -U '{FASTQ}'"
+                " -k 1 --mm -p {THREADS} {ALIGNER_ARGS}"
+            ),
             paired_cmd=(
                 "{BIN_PATH} -x '{INDEX_PATH}' -1 '{FASTQ1}' -2 '{FASTQ2}'"
-                " -k 1 --mm -p {THREADS}"
+                " -k 1 --mm -p {THREADS} {ALIGNER_ARGS}"
             ),
             idx_archive_fn="human-t2t-hla.tar",
             idx_name="human-t2t-hla",
@@ -58,8 +61,8 @@
             # cdn_base_url="http://localhost:8000",  # python -m http.server
             cdn_base_url=f"https://objectstorage.uk-london-1.oraclecloud.com/n/lrbvkel2wjot/b/human-genome-bucket/o",
             data_dir=XDG_DATA_DIR,
-            cmd="{BIN_PATH} -ax map-ont -m 40 --secondary no -t {THREADS} '{REF_ARCHIVE_PATH}' '{FASTQ}'",
-            paired_cmd="{BIN_PATH} -ax sr -m 40 --secondary no -t {THREADS} '{REF_ARCHIVE_PATH}' '{FASTQ1}' '{FASTQ2}'",
+            cmd="{BIN_PATH} -ax map-ont -m 40 --secondary no -t {THREADS} {ALIGNER_ARGS} '{REF_ARCHIVE_PATH}' '{FASTQ}'",
+            paired_cmd="{BIN_PATH} -ax sr -m 40 --secondary no -t {THREADS} {ALIGNER_ARGS} '{REF_ARCHIVE_PATH}' '{FASTQ1}' '{FASTQ2}'",
             ref_archive_fn="human-t2t-hla.fa.gz",
             idx_name="human-t2t-hla.fa.gz",
         ),
@@ -201,6 +204,7 @@ def clean_fastqs(
     sort_by_name: bool = False,
     out_dir: Path = CWD,
     aligner: ALIGNER = ALIGNER.minimap2,
+    aligner_args: str = "",
     threads: int = THREADS,
     force: bool = False,
 ):
@@ -220,6 +224,7 @@ def clean_fastqs(
             index=index,
             rename=rename,
             sort_by_name=sort_by_name,
+            aligner_args=aligner_args,
             threads=threads,
             force=force,
         )
@@ -242,6 +247,7 @@ def clean_paired_fastqs(
     sort_by_name: bool = False,
     out_dir: Path = CWD,
     aligner: ALIGNER = ALIGNER.bowtie2,
+    aligner_args: str = "",
     threads: int = THREADS,
     force: bool = False,
 ):
@@ -262,6 +268,7 @@ def clean_paired_fastqs(
             index=index,
             rename=rename,
             sort_by_name=sort_by_name,
+            aligner_args=aligner_args,
             threads=threads,
             force=force,
         )

tests/test_all.py

@@ -506,3 +506,37 @@ def test_paired_sort_rename():
     assert fifth_line_1 == "@2 /1"
     assert fifth_line_2 == "@2 /2"
     shutil.rmtree(out_dir, ignore_errors=True)
+
+
+def test_minimap2_aligner_args():
+    stats = lib.clean_fastqs(
+        fastqs=[data_dir / "sars-cov-2_100_1.fastq.gz"],
+        aligner=lib.ALIGNER.minimap2,
+        index=data_dir / "sars-cov-2/sars-cov-2.fasta.gz",
+        sort_by_name=True,
+        out_dir=out_dir,
+        aligner_args="-x asm5",  # Lets everything through
+        force=True,
+    )
+    assert stats[0]["reads_out"] == 50
+    shutil.rmtree(out_dir, ignore_errors=True)
+
+
+def test_bowtie2_aligner_args():
+    stats = lib.clean_paired_fastqs(
+        fastqs=[
+            (
+                data_dir / "sars-cov-2_100_1.fastq.gz",
+                data_dir / "sars-cov-2_100_2.fastq.gz",
+            ),
+        ],
+        aligner=lib.ALIGNER.bowtie2,
+        index=data_dir / "sars-cov-2/sars-cov-2",
+        rename=True,
+        sort_by_name=True,
+        out_dir=out_dir,
+        aligner_args="--ignore-quals",
+        force=True,
+    )
+    assert stats[0]["reads_out"] == 8
+    shutil.rmtree(out_dir, ignore_errors=True)