feat: hifiasm wrapper (#510)

* Added hifiasm wrapper

* Finished doc

* Added some files to keep memory usage down

* Update bio/hifiasm/environment.yaml

* Update test.py

Co-authored-by: Johannes Köster <johannes.koester@uni-due.de>

bio/hifiasm/environment.yaml

@@ -0,0 +1,6 @@
+channels:
+  - conda-forge
+  - bioconda
+  - nodefaults
+dependencies:
+  - hifiasm =0.16

bio/hifiasm/meta.yaml

@@ -0,0 +1,12 @@
+name: Hifiasm
+description: A haplotype-resolved assembler for accurate Hifi reads
+url: https://github.com/chhylp123/hifiasm
+authors:
+  - Filipe G. Vieira
+input:
+  - PacBio HiFi reads (fasta)
+  - Hi-C reads (fastq; optional)
+output:
+  - assembly graphs (GFA)
+notes: |
+  * The `extra` param allows for additional program arguments.

bio/hifiasm/test/Snakefile

@@ -0,0 +1,34 @@
+rule hifiasm:
+    input:
+        fasta=[
+            "reads/HiFi_dataset_01.fasta.gz",
+            "reads/HiFi_dataset_02.fasta.gz",
+        ],
+    # optional
+    # hic1="reads/Hi-C_dataset_R1.fastq.gz",
+    # hic2="reads/Hi-C_dataset_R2.fastq.gz",
+    output:
+        multiext(
+            "hifiasm/{sample}.",
+            "a_ctg.gfa",
+            "a_ctg.lowQ.bed",
+            "a_ctg.noseq.gfa",
+            "p_ctg.gfa",
+            "p_ctg.lowQ.bed",
+            "p_ctg.noseq.gfa",
+            "p_utg.gfa",
+            "p_utg.lowQ.bed",
+            "p_utg.noseq.gfa",
+            "r_utg.gfa",
+            "r_utg.lowQ.bed",
+            "r_utg.noseq.gfa",
+        ),
+    log:
+        "logs/hifiasm/{sample}.log",
+    params:
+        extra="--primary -f 37 -l 1 -s 0.75 -O 1",
+    threads: 2
+    resources:
+        mem_mb=1024,
+    wrapper:
+        "master/bio/hifiasm"

bio/hifiasm/wrapper.py

@@ -0,0 +1,38 @@
+__author__ = "Filipe G. Vieira"
+__copyright__ = "Copyright 2022, Filipe G. Vieira"
+__license__ = "MIT"
+
+
+import os
+from snakemake.shell import shell
+
+
+log = snakemake.log_fmt_shell()
+extra = snakemake.params.get("extra", "")
+
+
+hic1 = snakemake.input.get("hic1", "")
+if hic1:
+    if isinstance(hic1, list):
+        hic1 = ",".join(hic1)
+    hic1 = "--h1 {}".format(hic1)
+
+hic2 = snakemake.input.get("hic2", "")
+if hic2:
+    if isinstance(hic2, list):
+        hic2 = ",".join(hic2)
+    hic2 = "--h2 {}".format(hic2)
+
+
+out_prefix = os.path.commonprefix(snakemake.output).rstrip(".")
+
+
+shell(
+    "hifiasm"
+    " -t {snakemake.threads}"
+    " {extra}"
+    " {hic1} {hic2}"
+    " -o {out_prefix}"
+    " {snakemake.input.fasta}"
+    " {log}"
+)

test.py

@@ -140,6 +140,14 @@ def run(wrapper, cmd, check_log=None):
 
 
 
+@skip_if_not_modified
+def test_hifiasm():
+    run(
+        "bio/hifiasm",
+        ["snakemake", "--cores", "2", "hifiasm/a.a_ctg.gfa", "--use-conda", "-F"],
+    )
+
+
 @skip_if_not_modified
 def meryl_count():
     run(