Merge pull request #12 from nf-core/promoters

Add promoter prediction

conf/modules.config

@@ -28,6 +28,65 @@ process {
         ext.suffix = "bed"
     }
 
+    withName: FILTER_CONVERT_GTF {
+        ext.args = {"'BEGIN {OFS = \"\\t\"} \$3 == \"transcript\" {print \$1, \$4-1, \$4, \$1 \":\" \$4-1 \"-\" \$4, \$6, \$7}'"}
+        ext.suffix = "bed"
+    }
+
+    withName: SORT_BED {
+        ext.args = "-k1,1 -k2,2n"
+        ext.prefix = {"${meta.id}_sorted"}
+        ext.suffix = "bed"
+    }
+
+    withName: CONSTRUCT_TSS {
+        ext.args = "-b ${params.rose_tss_window}"
+        ext.prefix = "tss"
+    }
+
+    withName: FILTER_PREDICTIONS {
+        ext.args = "-A -f 1"
+        ext.prefix = {"${meta.id}_filtered"}
+    }
+
+    withName: STITCHING {
+        ext.args = "-d ${params.rose_stitching_window}"
+        ext.prefix = {"${meta.id}_stitched"}
+    }
+
+    withName: TSS_OVERLAP {
+        ext.args = "-c"
+        ext.prefix = {"${meta.id}_tss-overlap-counts"}
+    }
+
+    withName: FILTER_OVERLAPS {
+        ext.args = {"'BEGIN {OFS = \"\\t\"} \$NF >= 2 {print \$1, \$2, \$3}'"}
+        ext.prefix = {"${meta.id}_overlap"}
+    }
+
+    withName: UNSTITCHED_REGIONS {
+        ext.args = "-F 1"
+        ext.prefix = {"${meta.id}_original_regions"}
+    }
+
+    withName: CONCAT_AND_SORT {
+        ext.args = "-k1,1 -k2,2n"
+        ext.suffix = "bed"
+    }
+
+    withName: ".*:ROSE:CONCAT_AND_SORT" {
+        ext.prefix = {"${meta.id}_stitched"}
+    }
+
+    withName: ".*:FIMO:ADD_MISSING_COLUMNS" {
+        ext.args = "'BEGIN {OFS = \"\\t\"} {for (i = 1; i <= 6; i++) if (\$i == \"\") \$i = \".\"; print \$1, \$2, \$3, \$4, \$5, \$6}'"
+        ext.prefix = {"${meta.id}_unified"}
+    }
+
+    withName: ".*:FIMO:CONCAT_AND_SORT" {
+        ext.prefix = {"${meta.id}_sorted"}
+    }
+
     withName: BEDTOOLS_SORT {
         ext.prefix = {"${meta.id}.sorted"}
     }
@@ -54,18 +113,6 @@ process {
         ext.prefix = {"${meta.id}.merged"}
     }
 
-    withName: ".*:CHROMHMM:REHEADER.*" {
-        ext.args = "-c 'sed -e \"s/SN:\\([0-9XY]*\\)/SN:chr\\\\1/\" -e \"s/SN:MT/SN:chrM/\"'"
-    }
-
-    withName: ".*:CHROMHMM:REHEADER_CONTROL" {
-        ext.prefix = {"${meta.id}_control"}
-    }
-
-    withName: UCSC_GTFTOGENEPRED {
-        ext.args = "-genePredExt"
-    }
-
     withName: ".*DYNAMITE:FILTER" {
         ext.args = {"'BEGIN{OFS=\"\\t\"} NR==1 || (\$2 >= ${params.dynamite_min_regression} || \$2 <= -${params.dynamite_min_regression} )'"}
         ext.prefix = {"${meta.id}.filtered"}

main.nf

@@ -96,7 +96,8 @@ workflow NFCORE_TFACTIVITY {
         samplesheet_bam,
         params.chromhmm_states,
         params.chromhmm_threshold,
-        params.chromhmm_marks.split(','),
+        params.chromhmm_enhancer_marks.split(','),
+        params.chromhmm_promoter_marks.split(','),
 
         // Peaks
         params.window_size,

modules.json

@@ -10,16 +10,31 @@
                         "git_sha": "04bc484c987b523ea5420ed6bbc1fdc6d8aef751",
                         "installed_by": ["modules"]
                     },
+                    "bedtools/complement": {
+                        "branch": "master",
+                        "git_sha": "575e1bc54b083fb15e7dd8b5fcc40bea60e8ce83",
+                        "installed_by": ["modules"]
+                    },
                     "bedtools/getfasta": {
                         "branch": "master",
                         "git_sha": "cdcdd5e3d806f0ff3983c40c69e0b07bb44ec299",
                         "installed_by": ["modules"]
                     },
+                    "bedtools/intersect": {
+                        "branch": "master",
+                        "git_sha": "575e1bc54b083fb15e7dd8b5fcc40bea60e8ce83",
+                        "installed_by": ["modules"]
+                    },
                     "bedtools/merge": {
                         "branch": "master",
                         "git_sha": "a5377837fe9013bde89de8689829e83e84086536",
                         "installed_by": ["modules"]
                     },
+                    "bedtools/slop": {
+                        "branch": "master",
+                        "git_sha": "575e1bc54b083fb15e7dd8b5fcc40bea60e8ce83",
+                        "installed_by": ["modules"]
+                    },
                     "bedtools/sort": {
                         "branch": "master",
                         "git_sha": "571a5feac4c9ce0a8df0bc15b94230e7f3e8db47",
@@ -45,6 +60,11 @@
                         "git_sha": "cf3ed075695639b0a0924eb0901146df1996dc08",
                         "installed_by": ["modules"]
                     },
+                    "gnu/sort": {
+                        "branch": "master",
+                        "git_sha": "ca199cfe5aa4f1ea3c41302158f0af2cfaa58957",
+                        "installed_by": ["modules"]
+                    },
                     "gunzip": {
                         "branch": "master",
                         "git_sha": "3a5fef109d113b4997c9822198664ca5f2716208",
@@ -59,11 +79,6 @@
                         "branch": "master",
                         "git_sha": "04fbbc7c43cebc0b95d5b126f6d9fe4effa33519",
                         "installed_by": ["modules"]
-                    },
-                    "ucsc/gtftogenepred": {
-                        "branch": "master",
-                        "git_sha": "acb0880789a6ebc2168d3b2d3d42b5bce6a62431",
-                        "installed_by": ["modules"]
                     }
                 }
             },

modules/local/chromhmm/binarize_bams/main.nf

@@ -13,7 +13,8 @@ process BINARIZE_BAMS {
     tuple val(meta3), path(chromsizes)
 
     output:
-    tuple val(meta), path("output")
+    tuple val(meta), path("output"), emit: binarized_bams
+    path "versions.yml",             emit: versions
 
     script:
     """
@@ -22,5 +23,10 @@ process BINARIZE_BAMS {
         input \\
         $table \\
         output
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        chromhmm: \$(ChromHMM.sh Version | cut -f4 -d" ")
+    END_VERSIONS
     """
 }

modules/local/chromhmm/get_results/main.nf

@@ -13,9 +13,10 @@ process GET_RESULTS {
     val(marks)
 
     output:
-    tuple val(meta), path("$output_file")
+    tuple val(meta), path("$output_file"), emit: regions
+    path "versions.yml",                   emit: versions
 
     script:
-    output_file = "enhancers_${meta.id}.bed"
+    output_file = "${meta.id}.bed"
     template "get_results.py"
 }

modules/local/chromhmm/get_results/templates/get_results.py

@@ -3,6 +3,28 @@
 
 import pandas as pd
 import numpy as np
+import platform
+
+
+def format_yaml_like(data: dict, indent: int = 0) -> str:
+    """Formats a dictionary to a YAML-like string.
+
+    Args:
+        data (dict): The dictionary to format.
+        indent (int): The current indentation level.
+
+    Returns:
+        str: A string formatted as YAML.
+    """
+    yaml_str = ""
+    for key, value in data.items():
+        spaces = "  " * indent
+        if isinstance(value, dict):
+            yaml_str += f"{spaces}{key}:\\n{format_yaml_like(value, indent + 1)}"
+        else:
+            yaml_str += f"{spaces}{key}: {value}\\n"
+    return yaml_str
+
 
 marks = "${marks.join(' ')}".split()
 
@@ -30,3 +52,16 @@
 
 # Write output
 bed.to_csv("$output_file", index=False, sep="\\t", header=False)
+
+
+# Create version file
+versions = {
+    "${task.process}" : {
+        "python": platform.python_version(),
+        "pandas": pd.__version__,
+        "numpy": np.__version__,
+    }
+}
+
+with open("versions.yml", "w") as f:
+    f.write(format_yaml_like(versions))

modules/local/chromhmm/learn_model/main.nf

@@ -12,7 +12,8 @@ process LEARN_MODEL {
     val states
 
     output:
-    tuple val(meta), path("output/emissions_${states}.txt"), path("output/*_${states}_dense.bed")
+    tuple val(meta), path("output/emissions_${states}.txt"), path("output/*_${states}_dense.bed"), emit: model
+    path "versions.yml",                                                                           emit: versions
 
     script:
     """
@@ -24,5 +25,10 @@ process LEARN_MODEL {
         output \\
         $states \\
         PLACEHOLDER
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        chromhmm: \$(ChromHMM.sh Version | cut -f4 -d" ")
+    END_VERSIONS
     """
 }