sorting scripts depending on their function

loosolab · Jan 3, 2019 · b7c80c8 · b7c80c8
1 parent e29ad65
commit b7c80c8
Show file tree

Hide file tree

Showing 59 changed files with 5,771 additions and 9 deletions.
diff --git a/bin/footprints_extraction.py → ...print_extraction/footprints_extraction.py b/bin/footprints_extraction.py → ...print_extraction/footprints_extraction.py
diff --git a/bin/compareBed.sh → bin/1.2_filter_motifs/compareBed.sh b/bin/compareBed.sh → bin/1.2_filter_motifs/compareBed.sh
diff --git a/bin/maxScore.R → bin/1.2_filter_motifs/maxScore.R b/bin/maxScore.R → bin/1.2_filter_motifs/maxScore.R
diff --git a/bin/merge.R → bin/1.2_filter_motifs/merge.R b/bin/merge.R → bin/1.2_filter_motifs/merge.R
diff --git a/bin/tfbsscan.py → bin/1.2_filter_motifs/tfbsscan.py b/bin/tfbsscan.py → bin/1.2_filter_motifs/tfbsscan.py
diff --git a/bin/cdhit_wrapper.R → bin/2.1_clustering/cdhit_wrapper.R b/bin/cdhit_wrapper.R → bin/2.1_clustering/cdhit_wrapper.R
diff --git a/bin/reduce_sequence.R → bin/2.1_clustering/reduce_sequence.R b/bin/reduce_sequence.R → bin/2.1_clustering/reduce_sequence.R
diff --git a/bin/bed_to_fasta.R → bin/2.2_motif_estimation/bed_to_fasta.R b/bin/bed_to_fasta.R → bin/2.2_motif_estimation/bed_to_fasta.R
diff --git a/bin/get_best_motif.py → bin/2.2_motif_estimation/get_best_motif.py b/bin/get_best_motif.py → bin/2.2_motif_estimation/get_best_motif.py
diff --git a/bin/merge_similar_clusters.R → ...motif_estimation/merge_similar_clusters.R b/bin/merge_similar_clusters.R → ...motif_estimation/merge_similar_clusters.R
diff --git a/bin/motif_estimation.nf → bin/2.2_motif_estimation/motif_estimation.nf b/bin/motif_estimation.nf → bin/2.2_motif_estimation/motif_estimation.nf
diff --git a/bin/Modules/CrossMap.py → bin/3.1_create_gtf/Modules/CrossMap.py b/bin/Modules/CrossMap.py → bin/3.1_create_gtf/Modules/CrossMap.py
diff --git a/bin/Modules/CrossMapper.py → bin/3.1_create_gtf/Modules/CrossMapper.py b/bin/Modules/CrossMapper.py → bin/3.1_create_gtf/Modules/CrossMapper.py
diff --git a/bin/Modules/Ensembl/ActivityCategorizer.py → ...tf/Modules/Ensembl/ActivityCategorizer.py b/bin/Modules/Ensembl/ActivityCategorizer.py → ...tf/Modules/Ensembl/ActivityCategorizer.py
diff --git a/bin/Modules/Ensembl/ActivityTable.py → ...eate_gtf/Modules/Ensembl/ActivityTable.py b/bin/Modules/Ensembl/ActivityTable.py → ...eate_gtf/Modules/Ensembl/ActivityTable.py
diff --git a/...Modules/Ensembl/ActivityTableGenerator.py → ...Modules/Ensembl/ActivityTableGenerator.py b/...Modules/Ensembl/ActivityTableGenerator.py → ...Modules/Ensembl/ActivityTableGenerator.py
diff --git a/bin/Modules/Ensembl/Ensembl.py → ...3.1_create_gtf/Modules/Ensembl/Ensembl.py b/bin/Modules/Ensembl/Ensembl.py → ...3.1_create_gtf/Modules/Ensembl/Ensembl.py
diff --git a/bin/Modules/Ensembl/FTPHandling/FTPEntry.py → ...f/Modules/Ensembl/FTPHandling/FTPEntry.py b/bin/Modules/Ensembl/FTPHandling/FTPEntry.py → ...f/Modules/Ensembl/FTPHandling/FTPEntry.py
diff --git a/...odules/Ensembl/FTPHandling/URLRetrieve.py → ...odules/Ensembl/FTPHandling/URLRetrieve.py b/...odules/Ensembl/FTPHandling/URLRetrieve.py → ...odules/Ensembl/FTPHandling/URLRetrieve.py
diff --git a/...les/Ensembl/FTPHandling/VersionChecker.py → ...les/Ensembl/FTPHandling/VersionChecker.py b/...les/Ensembl/FTPHandling/VersionChecker.py → ...les/Ensembl/FTPHandling/VersionChecker.py
diff --git a/bin/Modules/Ensembl/FTPHandling/__init__.py → ...f/Modules/Ensembl/FTPHandling/__init__.py b/bin/Modules/Ensembl/FTPHandling/__init__.py → ...f/Modules/Ensembl/FTPHandling/__init__.py
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/FTPEntry.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/FTPEntry.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/URLRetrieve.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/URLRetrieve.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/VersionChecker.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/VersionChecker.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/__init__.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/FTPHandling/__pycache__/__init__.cpython-35.pyc
diff --git a/bin/Modules/Ensembl/GTFGen.py → bin/3.1_create_gtf/Modules/Ensembl/GTFGen.py b/bin/Modules/Ensembl/GTFGen.py → bin/3.1_create_gtf/Modules/Ensembl/GTFGen.py
diff --git a/bin/Modules/Ensembl/__init__.py → ....1_create_gtf/Modules/Ensembl/__init__.py b/bin/Modules/Ensembl/__init__.py → ....1_create_gtf/Modules/Ensembl/__init__.py
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/ActivityCategorizer.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/ActivityCategorizer.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/ActivityTable.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/ActivityTable.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/ActivityTableGenerator.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/ActivityTableGenerator.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/Ensembl.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/Ensembl.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/GTFGen.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/GTFGen.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/__init__.cpython-35.pyc b/bin/3.1_create_gtf/Modules/Ensembl/__pycache__/__init__.cpython-35.pyc
diff --git a/bin/Modules/Ensembl/checksums.py → ...1_create_gtf/Modules/Ensembl/checksums.py b/bin/Modules/Ensembl/checksums.py → ...1_create_gtf/Modules/Ensembl/checksums.py
diff --git a/bin/Modules/SaveResults.py → bin/3.1_create_gtf/Modules/SaveResults.py b/bin/Modules/SaveResults.py → bin/3.1_create_gtf/Modules/SaveResults.py
diff --git a/bin/Modules/Uniquifier.py → bin/3.1_create_gtf/Modules/Uniquifier.py b/bin/Modules/Uniquifier.py → bin/3.1_create_gtf/Modules/Uniquifier.py
diff --git a/bin/Modules/__init__.py → bin/3.1_create_gtf/Modules/__init__.py b/bin/Modules/__init__.py → bin/3.1_create_gtf/Modules/__init__.py
diff --git a/bin/3.1_create_gtf/Modules/__pycache__/CrossMap.cpython-35.pyc b/bin/3.1_create_gtf/Modules/__pycache__/CrossMap.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/__pycache__/CrossMapper.cpython-35.pyc b/bin/3.1_create_gtf/Modules/__pycache__/CrossMapper.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/__pycache__/SaveResults.cpython-35.pyc b/bin/3.1_create_gtf/Modules/__pycache__/SaveResults.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/__pycache__/Uniquifier.cpython-35.pyc b/bin/3.1_create_gtf/Modules/__pycache__/Uniquifier.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/__pycache__/__init__.cpython-35.pyc b/bin/3.1_create_gtf/Modules/__pycache__/__init__.cpython-35.pyc
diff --git a/bin/Modules/ucsc/__init__.py → bin/3.1_create_gtf/Modules/ucsc/__init__.py b/bin/Modules/ucsc/__init__.py → bin/3.1_create_gtf/Modules/ucsc/__init__.py
diff --git a/bin/3.1_create_gtf/Modules/ucsc/__pycache__/__init__.cpython-35.pyc b/bin/3.1_create_gtf/Modules/ucsc/__pycache__/__init__.cpython-35.pyc
diff --git a/bin/3.1_create_gtf/Modules/ucsc/__pycache__/ucsc.cpython-35.pyc b/bin/3.1_create_gtf/Modules/ucsc/__pycache__/ucsc.cpython-35.pyc
diff --git a/bin/Modules/ucsc/bigBedToBed → bin/3.1_create_gtf/Modules/ucsc/bigBedToBed b/bin/Modules/ucsc/bigBedToBed → bin/3.1_create_gtf/Modules/ucsc/bigBedToBed
diff --git a/bin/Modules/ucsc/ucsc.py → bin/3.1_create_gtf/Modules/ucsc/ucsc.py b/bin/Modules/ucsc/ucsc.py → bin/3.1_create_gtf/Modules/ucsc/ucsc.py
diff --git a/bin/RegGTFExtractor.py → bin/3.1_create_gtf/RegGTFExtractor.py b/bin/RegGTFExtractor.py → bin/3.1_create_gtf/RegGTFExtractor.py
diff --git a/bin/config/celltypes_homo_sapiens.json → ...te_gtf/config/celltypes_homo_sapiens.json b/bin/config/celltypes_homo_sapiens.json → ...te_gtf/config/celltypes_homo_sapiens.json
diff --git a/bin/config/celltypes_mus_musculus.json → ...te_gtf/config/celltypes_mus_musculus.json b/bin/config/celltypes_mus_musculus.json → ...te_gtf/config/celltypes_mus_musculus.json
diff --git a/bin/3.1_create_gtf/data/EnsemblData/release-94/homo_sapiens/CHECKSUMS b/bin/3.1_create_gtf/data/EnsemblData/release-94/homo_sapiens/CHECKSUMS
@@ -0,0 +1,3 @@
+fd500a8b924f0f3ddc5391baf259b429  homo_sapiens.GRCh38.motiffeatures.20161111.gff.gz
+b42e919ff359bd873c7e5eea14b49776  homo_sapiens.GRCh38.Regulatory_Build.regulatory_features.20161111.gff.gz
+dbd442cdd993ca44cbbf39be620dfa23  README
diff --git a/bin/3.1_create_gtf/data/EnsemblData/release-94/homo_sapiens/activity/A549/CHECKSUMS b/bin/3.1_create_gtf/data/EnsemblData/release-94/homo_sapiens/activity/A549/CHECKSUMS
@@ -0,0 +1 @@
+10483  8961 homo_sapiens.GRCh38.A549.Regulatory_Build.regulatory_activity.20161111.gff.gz
diff --git a/...tivity/A549/homo_sapiens.GRCh38.A549.Regulatory_Build.regulatory_activity.20161111.gff.gz b/...tivity/A549/homo_sapiens.GRCh38.A549.Regulatory_Build.regulatory_activity.20161111.gff.gz
diff --git a/bin/3.1_create_gtf/data/EnsemblData/release-94/homo_sapiens/activity/Aorta/CHECKSUMS b/bin/3.1_create_gtf/data/EnsemblData/release-94/homo_sapiens/activity/Aorta/CHECKSUMS
@@ -0,0 +1 @@
+37136  8883 homo_sapiens.GRCh38.Aorta.Regulatory_Build.regulatory_activity.20161111.gff.gz
diff --git a/...vity/Aorta/homo_sapiens.GRCh38.Aorta.Regulatory_Build.regulatory_activity.20161111.gff.gz b/...vity/Aorta/homo_sapiens.GRCh38.Aorta.Regulatory_Build.regulatory_activity.20161111.gff.gz
diff --git a/...-94/homo_sapiens/homo_sapiens.GRCh38.Regulatory_Build.regulatory_features.20161111.gff.gz b/...-94/homo_sapiens/homo_sapiens.GRCh38.Regulatory_Build.regulatory_features.20161111.gff.gz
diff --git a/bin/3.1_create_gtf/data/UCSCData/hg38.bed b/bin/3.1_create_gtf/data/UCSCData/hg38.bed
diff --git a/masterenv.yml b/masterenv.yml
@@ -1,6 +1,7 @@
 
 name: masterenv
 dependencies:
+  - python >=3
   - r-seqinr
   - numpy
   - pybigWig

diff --git a/pipeline.nf b/pipeline.nf
@@ -203,7 +203,7 @@ process footprint_extraction {
 
 	script:
 	"""
-	python ${path_bin}/footprints_extraction.py --bigwig ${bigWig} --bed ${bed} --output_file ${name}_called_peaks.bed --window_length ${params.window_length} --step ${params.step} --percentage ${params.percentage}
+	python ${path_bin}/1.1_footprint_extraction/footprints_extraction.py --bigwig ${bigWig} --bed ${bed} --output_file ${name}_called_peaks.bed --window_length ${params.window_length} --step ${params.step} --percentage ${params.percentage}
 	"""
 }
 
@@ -229,7 +229,7 @@ process extract_known_TFBS {
 
 	script:
 	"""
-	python ${path_bin}/tfbsscan.py --use moods --core ${params.threads} -m ${db} -g ${fasta} -o ${params.create_known_tfbs_path} -b ${bed}
+	python ${path_bin}/1.2_filter_motifs/tfbsscan.py --use moods --core ${params.threads} -m ${db} -g ${fasta} -o ${params.create_known_tfbs_path} -b ${bed}
 	"""
 }
 
@@ -263,7 +263,7 @@ process overlap_with_known_TFBS {
 	script:
 	motif_list = bed_motifs.toString().replaceAll(/\s|\[|\]/,"")
 	"""
-	${path_bin}/compareBed.sh --data ${bed_footprints} --motifs ${motif_path} --fasta ${fasta} -o ${name}_unknown.bed -min ${params.min_size_fp} -max ${params.max_size_fp} -p ${path_bin}
+	${path_bin}/1.2_filter_motifs/compareBed.sh --data ${bed_footprints} --motifs ${motif_path} --fasta ${fasta} -o ${name}_unknown.bed -min ${params.min_size_fp} -max ${params.max_size_fp} -p ${path_bin}/1.2/filter_motifs
 	"""
 }
 
@@ -284,7 +284,7 @@ process reduce_sequence {
 
 	script:
 	"""
-	Rscript ${path_bin}/reduce_sequence.R -i ${bed} -k ${params.kmer} -m ${params.aprox_motif_len} -o ${name}_reduced.bed -t ${params.threads} -f ${params.motif_occurence} -s ${params.min_seq_length}
+	Rscript ${path_bin}/2.1_clustering/reduce_sequence.R -i ${bed} -k ${params.kmer} -m ${params.aprox_motif_len} -o ${name}_reduced.bed -t ${params.threads} -f ${params.motif_occurence} -s ${params.min_seq_length}
 	"""
 }
 
@@ -306,7 +306,7 @@ process clustering {
 
 	script:
 	"""
-	Rscript ${path_bin}/cdhit_wrapper.R -i ${bed} -A ${params.sequence_coverage} -o ${name}_clusterd.bed -c ${params.identity} -G ${params.global} -M ${params.memory} -l ${params.throw_away_seq} -r ${params.strand} -T ${params.threads}
+	Rscript ${path_bin}/2.1_clustering/cdhit_wrapper.R -i ${bed} -A ${params.sequence_coverage} -o ${name}_clusterd.bed -c ${params.identity} -G ${params.global} -M ${params.memory} -l ${params.throw_away_seq} -r ${params.strand} -T ${params.threads}
 	"""
 }
 
@@ -328,7 +328,7 @@ process bed_to_clustered_fasta {
 
     script:
     """
-    Rscript ${path_bin}/bed_to_fasta.R -i ${bed} -p ${name} -m ${params.min_seq}
+    Rscript ${path_bin}/2.2_motif_estimation/bed_to_fasta.R -i ${bed} -p ${name} -m ${params.min_seq}
     """
 }
 
@@ -436,7 +436,7 @@ process merge_fasta {
   	fa_sorted = fasta_list.sort(false) { it.getBaseName().tokenize('_')[-1] as Integer }
   	fastalist = fa_sorted.toString().replaceAll(/\s|\[|\]/,"")
 	"""
-	Rscript ${path_bin}/merge_similar_clusters.R ${motiv_sim} ${fastalist} ${params.edge_weight}
+	Rscript ${path_bin}/2.2_motif_estimation/merge_similar_clusters.R ${motiv_sim} ${fastalist} ${params.edge_weight}
 	"""
 }
 
@@ -539,7 +539,7 @@ process get_best_motif {
 
 	script:
 	"""
-	python ${path_bin}/get_best_motif.py ${meme} ${name}_best.meme ${params.best_motif}
+	python ${path_bin}/2.2_motif_estimation/get_best_motif.py ${meme} ${name}_best.meme ${params.best_motif}
 	"""
 }
 
@@ -589,7 +589,7 @@ process create_GTF {
 
 	script:
 	"""
-	python ${path_bin}/RegGTFExtractor.py ${params.organism} --tissue ${params.tissues} --wd ${path_bin}
+	python ${path_bin}/3.1_create_gtf/RegGTFExtractor.py ${params.organism} --tissue ${params.tissues} --wd ${path_bin}/3.1_create_gtf/
 	"""
 }
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		10483 8961 homo_sapiens.GRCh38.A549.Regulatory_Build.regulatory_activity.20161111.gff.gz
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		37136 8883 homo_sapiens.GRCh38.Aorta.Regulatory_Build.regulatory_activity.20161111.gff.gz