added per sample fasta rotation

metagenlab · Oct 24, 2024 · e9a0311 · e9a0311
1 parent f987849
commit e9a0311
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 27 deletions.
diff --git a/mess/workflow/rules/preflight/functions.smk b/mess/workflow/rules/preflight/functions.smk
@@ -114,24 +114,6 @@ def get_value(value, wildcards):
     return df.loc[vals, value]
 
 
-fa_table_cache = {}
-def get_random_start(wildcards):
-    table = checkpoints.split_contigs.get(**wildcards).output[0]
-    if table not in fa_table_cache:
-        df = pd.read_csv(
-            table,
-            sep="\t",
-            index_col=["fasta","contig","n"]
-        )
-        fa_table_cache[table] = df
-    vals=(f"{wildcards.fasta}",f"{wildcards.contig}",int(wildcards.n))
-    df = fa_table_cache[table]
-    return df.loc[vals, "random_start"]
-
-
-
-
-
 def get_asm_summary(wildcards):
     try:
         table = checkpoints.download_assemblies.get(**wildcards).output[0]

diff --git a/mess/workflow/rules/processing/fastas.smk b/mess/workflow/rules/processing/fastas.smk
@@ -64,17 +64,19 @@ if ROTATE > 1:
 
     rule rotate_contigs:
         input:
-            fa=os.path.join(dir.out.processing, "split", "{fasta}_{contig}.fna"),
-            df=get_cov_table,
+            os.path.join(dir.out.processing, "split", "{fasta}_{contig}.fna"),
         output:
-            os.path.join(dir.out.processing, "rotate", "{fasta}_{contig}_{n}.fna"),
+            os.path.join(
+                dir.out.processing, "rotate", "{sample}", "{fasta}_{contig}_{n}.fna"
+            ),
         params:
-            get_random_start,
+            lambda wildcards: get_value("random_start", wildcards),
         log:
             os.path.join(
                 dir.out.logs,
                 "seqkit",
                 "restart",
+                "{sample}",
                 "{fasta}_{contig}_{n}.log",
             ),
         resources:
@@ -88,7 +90,6 @@ if ROTATE > 1:
             containers.seqkit
         shell:
             """
-            seqkit restart -i {params} {input.fa} | \
-            seqkit seq -i | \
+            seqkit restart -i {params} {input} | \\
             seqkit replace -p .+ -r {wildcards.contig}_{wildcards.n} > {output}
             """
diff --git a/mess/workflow/rules/simulate/long_reads.smk b/mess/workflow/rules/simulate/long_reads.smk
@@ -2,7 +2,9 @@ prefix = os.path.join(dir.out.long, "{sample}", "{fasta}", "{contig}")
 fasta = os.path.join(dir.out.processing, "split", "{fasta}_{contig}.fna")
 if ROTATE > 1:
     prefix = os.path.join(dir.out.long, "{sample}", "{fasta}", "{contig}_{n}")
-    fasta = os.path.join(dir.out.processing, "rotate", "{fasta}_{contig}_{n}.fna")
+    fasta = os.path.join(
+        dir.out.processing, "rotate", "{sample}", "{fasta}_{contig}_{n}.fna"
+    )
 id_prefix = os.path.basename(prefix)
 
 if PASSES > 1:

diff --git a/mess/workflow/rules/simulate/short_reads.smk b/mess/workflow/rules/simulate/short_reads.smk
@@ -38,7 +38,9 @@ if not PAIRED:
 
 fasta = os.path.join(dir.out.processing, "split", "{fasta}_{contig}.fna")
 if ROTATE > 1:
-    fasta = os.path.join(dir.out.processing, "rotate", "{fasta}_{contig}_{n}.fna")
+    fasta = os.path.join(
+        dir.out.processing, "rotate", "{sample}", "{fasta}_{contig}_{n}.fna"
+    )
 
 
 rule art_illumina:
@@ -56,7 +58,7 @@ rule art_illumina:
         prefix=fq_prefix,
     log:
         os.path.join(dir.out.logs, "art", "{sample}", "{fasta}", "{contig}.log")
-        if ROTATE == 1
+        if not ROTATE == 1
         else os.path.join(
             dir.out.logs, "art", "{sample}", "{fasta}", "{contig}_{n}.log"
         ),