Fix missing rule all for PIQ and motif dictionary

ParkerLab · Sep 26, 2019 · 1c1cb02 · 1c1cb02
1 parent f67d36f
commit 1c1cb02
Show file tree

Hide file tree

Showing 4 changed files with 49 additions and 22 deletions.
diff --git a/analyses/1e_bmo/Snakefile b/analyses/1e_bmo/Snakefile
@@ -108,6 +108,21 @@ VERSION_DIR = join(RESULTS, "versions")
 # will be used to determine the output files for each sample
 subsample_depths = {x: subsampled_depths_from_sample(x) for x in get_samples()}
 
+# Number of motifs and motif-number dictionary - used for PIQ
+# e.g. 9-CTCFknown2, 20-NFKBknown10
+# Wildcards
+motifs = get_motifs()
+motifs = list(set(motifs)) # remove duplicates
+motifs.sort()
+nmotifs = len(motifs)
+motif_dic = {}
+motif_handle_dic = {}
+for i in range(0, nmotifs):
+    motif = motifs[i]
+    handle = "{}-{}".format(i + 1, motifs[i].replace("_", "").replace(".", ""))
+    motif_dic[motif] = str(i + 1)
+    motif_handle_dic[motif] = handle
+
 
 rule all:
     input:
@@ -180,7 +195,7 @@ rule all:
             ),
             sample=sample,
             depth=depths,
-            motif=get_motifs()
+            motif=motifs
         ) for sample, depths in subsample_depths.items()],
         # PIQ - scores files
         [expand(
@@ -190,7 +205,7 @@ rule all:
             ),
             sample=sample,
             depth=depths,
-            motif=get_motifs()
+            motif=motifs
         ) for sample, depths in subsample_depths.items()],
 
         # Motifs in peaks (binary)

diff --git a/analyses/1e_bmo/rules/tf-binding_piq.smk b/analyses/1e_bmo/rules/tf-binding_piq.smk
@@ -31,7 +31,7 @@ rule gather_motifs:
     input:
         expand(
             os.path.join(PIQ_DIR, "motifs", "{motif}.jaspar"),
-            motif=get_motifs()
+            motif=motifs
         )
     output:
         mat = os.path.join(PIQ_DIR, "motifs", "matrices.txt")
@@ -42,19 +42,8 @@ rule gather_motifs:
         cat {input} > {output.mat}
         """
 
-rule pwm_key:
-    input:
-        rules.gather_motifs.output,
-    output:
-        os.path.join(PIQ_DIR, "motifs", "key.txt")
-    shell:
-        """
-        grep "^>" {input} | sed 's/^>//g' > tmp.keys.txt
-        sed 's/_//g' tmp.keys.txt > tmp.keys2.txt
-        paste tmp.keys.txt tmp.keys2.txt | cat -n > {output} && \
-        rm tmp.keys*txt
-        """
-
+# Note that variable i is defined in rule piq, but snakemake 
+# is able to propagate it bottom-up
 rule pwm_scan:
     input:
         mat = os.path.join(PIQ_DIR, "motifs", "matrices.txt"),

diff --git a/analyses/1e_bmo/rules_hepg2/Snakefile.smk b/analyses/1e_bmo/rules_hepg2/Snakefile.smk
@@ -170,14 +170,15 @@ subsample_depths = {x: subsampled_depths_from_sample(x) for x in get_samples()}
 
 # Number of motifs and motif-number dictionary - used for PIQ
 # e.g. 9-CTCFknown2, 20-NFKBknown10
-all_motifs = [x for x in get_motifs()]
-all_motifs.sort()
-nmotifs = len(all_motifs)
+motifs = get_motifs()
+motifs = list(set(motifs)) # remove duplicates
+motifs.sort()
+nmotifs = len(motifs)
 motif_dic = {}
 motif_handle_dic = {}
 for i in range(0, nmotifs):
-    motif = all_motifs[i]
-    handle = "{}-{}".format(i + 1, all_motifs[i].replace("_", ""))
+    motif = motifs[i]
+    handle = "{}-{}".format(i + 1, motifs[i].replace("_", "").replace(".", ""))
     motif_dic[motif] = str(i + 1)
     motif_handle_dic[motif] = handle
 
@@ -240,6 +241,26 @@ rule all:
             depth=depths,
             motif=get_motifs()
         ) for sample, depths in subsample_depths.items()],
+        # PIQ - binarized files
+        [expand(
+            os.path.join(
+                PIQ_DIR,
+                "processed_output_binary", "{sample}", "{depth}M.{motif}.bed"
+            ),
+            sample=sample,
+            depth=depths,
+            motif=motifs
+        ) for sample, depths in subsample_depths.items()],
+        # PIQ - scores files
+        [expand(
+            os.path.join(
+                PIQ_DIR,
+                "processed_output_scores", "{sample}", "{depth}M.{motif}.bed"
+            ),
+            sample=sample,
+            depth=depths,
+            motif=motifs
+        ) for sample, depths in subsample_depths.items()],
         # Motifs in peaks
         [expand(
             os.path.join(PEAKS_DIR, "{sample}", "{motif}.{depth}.bed"),

diff --git a/analyses/1e_bmo/rules_hepg2/tf-binding_piq.smk b/analyses/1e_bmo/rules_hepg2/tf-binding_piq.smk
@@ -31,7 +31,7 @@ rule gather_motifs:
     input:
         expand(
             os.path.join(PIQ_DIR, "motifs", "{motif}.jaspar"),
-            motif=get_motifs()
+            motif=motifs
         )
     output:
         mat = os.path.join(PIQ_DIR, "motifs", "matrices.txt")
@@ -55,6 +55,8 @@ rule pwm_key:
         rm tmp.keys*txt
         """
 
+# Note that variable i is defined in rule piq, but snakemake 
+# is able to propagate it bottom-up
 rule pwm_scan:
     input:
         mat = os.path.join(PIQ_DIR, "motifs", "matrices.txt"),