issue #72: added wtdbg2 (LR assembler)

workflow/envs/wtdbg2.yaml

+name: perl
+channels:
+  - anaconda
+  - conda-forge
+  - bioconda
+  - defaults
+  - r
+  - etetoolkit
+  - au-eoed
+dependencies:
+  - java-jdk=7.0.91
+  - libgcc-ng=9.1.0
+  - perl=5.26.2
+

workflow/rules/assembly.smk

@@ -7,7 +7,6 @@ rule assembly_lr_flye:
     input:
         os.path.join(RESULTS_DIR, "preproc/metag/lr/lr.fastq.gz")
     output:
-        # NOTE: NOT the name for "final" FASTA file which will be created AFTER polishing
         os.path.join(RESULTS_DIR, "assembly/lr/flye/assembly.fasta")
     threads:
         config["flye"]["threads"]
@@ -70,9 +69,9 @@ rule polishing_lr_medaka:
         # "cd $(dirname {output}) && ln -sf consensus.fasta $(basename {output}) && " # NOTE: flye already creates this file
         "date) &> {log}"
 
-localrules: assembly_lr_flye_link
+localrules: assembly_lr_link
 
-rule assembly_lr_flye_link:
+rule assembly_lr_link:
     input:
         os.path.join(RESULTS_DIR, "assembly/lr/{tool}/polishing/racon_medaka/consensus.fasta")
     output:
@@ -80,6 +79,37 @@ rule assembly_lr_flye_link:
     shell:
         "ipath=$(realpath {input}) && cd $(dirname {output}) && ln -sf $(realpath --relative-to=\".\" \"${{ipath}}\") $(basename {output})"
 
+rule assembly_lr_wtdbg2:
+    input:
+        os.path.join(RESULTS_DIR, "preproc/metag/lr/lr.fastq.gz")
+    output:
+        os.path.join(RESULTS_DIR, "assembly/lr/wtdbg2/assembly.fasta")
+    threads:
+        config["wtdbg2"]["threads"]
+    log:
+        "logs/wtdbg2.log"
+    conda:
+        os.path.join(ENV_DIR, "wtdbg2.yaml")
+    message:
+        "Assembly: long reads: wtdbg2"
+    shell:
+        "(date && "
+        "export PATH=$PATH:{config[wtdbg2][bin]} && "
+        "wtdbg2 -x ont -g {config[wtdbg2][genome_size]} -t {threads} -i {input} -fo $(dirname {output})/assembly && "
+        "wtpoa-cns -t {threads} -i $(dirname {output})/assembly.ctg.lay.gz -fo {output} && "
+        "date) &> {log}"
+
+# localrules: assembly_lr_wtdbg2_link
+
+# NOTE: only for testing !!!
+# rule assembly_lr_wtdbg2_link:
+#     input:
+#         os.path.join(RESULTS_DIR, "assembly/lr/wtdbg2/assembly.fasta")
+#     output:
+#         os.path.join(RESULTS_DIR, "assembly/lr/wtdbg2/ASSEMBLY.fasta")
+#     shell:
+#         "ipath=$(realpath {input}) && cd $(dirname {output}) && ln -sf $(realpath --relative-to=\".\" \"${{ipath}}\") $(basename {output})"
+
 ##################################################
 # Short reads
 

workflow/steps/assembly.smk

@@ -7,8 +7,10 @@ include:
 
 rule ASSEMBLY:
     input:
-       lr=expand(os.path.join(RESULTS_DIR, "assembly/lr/{tool}/ASSEMBLY.FILTERED.fasta"), tool=config["assemblers"]["lr"]),
-       sr=expand(os.path.join(RESULTS_DIR, "assembly/sr/{tool}/ASSEMBLY.FILTERED.fasta"), tool=config["assemblers"]["sr"]),
-       hy=expand(os.path.join(RESULTS_DIR, "assembly/hy/{tool}/ASSEMBLY.FILTERED.fasta"), tool=config["assemblers"]["hy"]),
+    #    lr=expand(os.path.join(RESULTS_DIR, "assembly/lr/{tool}/ASSEMBLY.FILTERED.fasta"), tool=config["assemblers"]["lr"]),
+        lr=expand(os.path.join(RESULTS_DIR, "assembly/lr/{tool}/ASSEMBLY.FILTERED.fasta"), tool=["flye"]),
+        lr2=expand(os.path.join(RESULTS_DIR, "assembly/lr/{tool}/ASSEMBLY.fasta"), tool=["wtdbg2"]),
+        sr=expand(os.path.join(RESULTS_DIR, "assembly/sr/{tool}/ASSEMBLY.FILTERED.fasta"), tool=config["assemblers"]["sr"]),
+        hy=expand(os.path.join(RESULTS_DIR, "assembly/hy/{tool}/ASSEMBLY.FILTERED.fasta"), tool=config["assemblers"]["hy"]),
     output:
         touch("status/assembly.done")