Merge pull request #247 from CDCgov/rjd0-val-params

Rjd0 val params
CDCgov · Jan 11, 2025 · 4c6846a · 4c6846a
2 parents 3fc1971 + 6fc0b89
commit 4c6846a
Show file tree

Hide file tree

Showing 5 changed files with 37 additions and 177 deletions.
diff --git a/README.md b/README.md
@@ -7,6 +7,10 @@
 
 For the complete TOSTADAS documentation, please see the [Wiki](https://github.com/CDCgov/tostadas/wiki)
 
+## Warnings
+### Plugin Compatibility Warning
+❗ Important Note: This pipeline uses the nf-schema plugin to validate pipeline parameters. Users with Nextflow version 24 or later may encounter a warning message indicating that the plugin must be installed. To resolve this warning message, please install the plugin manually by following the instructions found in this [link](https://www.nextflow.io/docs/latest/plugins.html#offline-usage)
+
 ## Overview
 **T O S T A D A S**  
 **T**oolkit for **O**pen **S**equence **T**riage, **A**nnotation, and **DA**tabase **S**ubmission  
@@ -41,28 +45,17 @@ bash Mambaforge-$(uname)-$(uname -m).sh -b -p $HOME/mambaforge
 ```
 export PATH="$HOME/mambaforge/bin:$PATH"
 ```
-### 3. Create and activate a conda environment
-
- **3a. Create an empty conda environment**
-```
-conda create --name tostadas
-```
-This conda environment will be used to install Nextflow.
-
- **3b. Activate the environment**
-```
-conda activate tostadas
-```
-Verify which environment is active by running the following conda command: `conda env list`. The active environment will be denoted with an asterisk *
-
-### 4. Install Nextflow using mamba and the bioconda Channel
+### 3. Install Nextflow using mamba and the bioconda Channel
 ```
 mamba install -c bioconda nextflow
 ```
-### 5. Update the default submissions config file with your NCBI username and password, and run the following nextflow command to execute the scripts with default parameters and the local run environment: 
+### 4. Update the default submissions config file with your NCBI username and password 
 ```
 # update this config file (you don't have to use vim)
 vim conf/submission_config.yaml
+```
+### 5. Run the workflow with default parameters and the local run environment:
+```
 # test command for virus reads
 nextflow run main.nf -profile test,<singularity|docker|conda> --virus
 ```

diff --git a/modules/local/general_util/validate_params/main.nf b/modules/local/general_util/validate_params/main.nf
diff --git a/modules/local/general_util/validate_params/update-function.nf b/modules/local/general_util/validate_params/update-function.nf
diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -10,19 +10,20 @@
       "format": "file-path",
       "description": "Meta-data file path for samples",
       "default": "${projectDir}/assets/metadata_template.xlsx",
-      "pattern": "^\\S+\\.xlsx$"
+      "pattern": "^\\S+\\.(xlsx|xls|xlsm|xlsb|csv)$"
     },
     "ref_fasta_path": {
       "type": "string",
       "format": "file-path",
       "description": "Reference Sequence file path",
-      "default": "${projectDir}/assets/ref/ref.MPXV.NC063383.v7.fasta"
+      "default": "${projectDir}/assets/ref/ref.MPXV.NC063383.v7.fasta",
+      "pattern": "^\\S+\\.(fasta|fa|fna|ffn|faa|frn|fasta\\.gz|fa\\.gz|fna\\.gz|ffn\\.gz|faa\\.gz|frn\\.gz|fasta\\.zip|fa\\.zip|fna\\.zip|ffn\\.zip|faa\\.zip|frn\\.zip|fasta\\.bz2|fa\\.bz2|fna\\.bz2|ffn\\.bz2|faa\\.bz2|frn\\.bz2|fasta\\.tar\\.bz2|fa\\.tar\\.bz2|fna\\.tar\\.bz2|ffn\\.tar\\.bz2|faa\\.tar\\.bz2|frn\\.tar\\.bz2|fasta\\.tar|fa\\.tar|fna\\.tar|ffn\\.tar|faa\\.tar|frn\\.tar|fasta\\.tar\\.gz|fa\\.tar\\.gz|fna\\.tar\\.gz|ffn\\.tar\\.gz|faa\\.tar\\.gz|frn\\.tar\\.gz)$"
     },
     "ref_gff_path": {
       "type": "string",
       "format": "file-path",
-      "description": "Reference gff file path for annotation",
-      "default": "${projectDir}/assets/ref/ref.MPXV.NC063383.v7.gff"
+      "default": "${projectDir}/assets/ref/ref.MPXV.NC063383.v7.gff",
+      "pattern": "^\\S+\\.(gff|gff3)$"
     },
     "output_dir": {
       "type": "string",
@@ -56,8 +57,8 @@
     },
     "custom_fields_file": {
       "type": "string",
-      "description": "Path to the JSON file containing custom metadata fields and their information",
-      "default": "${projectDir}/assets/custom_meta_fields/example_custom_fields.json"
+      "default": "${projectDir}/assets/custom_meta_fields/example_custom_fields.json",
+      "pattern": "^\\S+\\.json$"
     },
     "final_liftoff_output_dir": {
       "type": "string",
@@ -90,7 +91,8 @@
     "lift_unmapped_features_file_name": {
       "type": "string",
       "description": "Name of unmapped features file name",
-      "default": "output.unmapped_features.txt"
+      "default": "output.unmapped_features.txt",
+      "pattern": "^\\S+\\.txt$"
     },
     "lift_copy_threshold": {
       "type": "number",
@@ -249,7 +251,8 @@
     "submission_config": {
       "type": "string",
       "description": "Configuration file for submission to public repos",
-      "default": "${projectDir}/bin/config_files/<your-ncbi-config>.yaml"
+      "default": "${projectDir}/bin/config_files/<your-ncbi-config>.yaml",
+      "pattern": "^\\S+\\.(yaml|yml)$"
     },
     "submission_wait_time": {
       "type": "integer",
@@ -297,7 +300,8 @@
     "repeat_library": {
       "type": "string",
       "description": "Path to the repeat library file used by RepeatMasker.",
-      "default": "${projectDir}/assets/lib/MPOX_repeats_lib.fasta"
+      "default": "${projectDir}/assets/lib/MPOX_repeats_lib.fasta",
+      "pattern": "^\\S+\\.(fasta|fa|fna|ffn|faa|frn|fasta\\.gz|fa\\.gz|fna\\.gz|ffn\\.gz|faa\\.gz|frn\\.gz|fasta\\.zip|fa\\.zip|fna\\.zip|ffn\\.zip|faa\\.zip|frn\\.zip|fasta\\.bz2|fa\\.bz2|fna\\.bz2|ffn\\.bz2|faa\\.bz2|frn\\.bz2|fasta\\.tar\\.bz2|fa\\.tar\\.bz2|fna\\.tar\\.bz2|ffn\\.tar\\.bz2|faa\\.tar\\.bz2|frn\\.tar\\.bz2|fasta\\.tar|fa\\.tar|fna\\.tar|ffn\\.tar|faa\\.tar|frn\\.tar|fasta\\.tar\\.gz|fa\\.tar\\.gz|fna\\.tar\\.gz|ffn\\.tar\\.gz|faa\\.tar\\.gz|frn\\.tar\\.gz)$"
     },
     "gisaid": {
       "description": "Flag to enable or disable submission to GISAID."
@@ -383,7 +387,8 @@
     },
     "lift_feature_types": {
       "type": "string",
-      "default": "${projectDir}/assets/feature_types.txt"
+      "default": "${projectDir}/assets/feature_types.txt",
+      "pattern": "^\\S+\\.txt$"
     },
     "processed_samples": {
       "type": "string",
@@ -429,7 +434,10 @@
       "default": "${projectDir}/environment.yml"
     }
   },
-
-  "required": ["meta_path", "ref_fasta_path", "ref_gff_path", "species"]
+  "required": [
+    "meta_path",
+    "ref_fasta_path",
+    "ref_gff_path",
+    "species"
+  ]
 }
-
diff --git a/workflows/tostadas.nf b/workflows/tostadas.nf
@@ -9,7 +9,7 @@ nextflow.enable.dsl=2
 // get the utility processes / subworkflows
 // include { CHECK_FILES                                       } from "../modules/local/general_util/check_files/main"
 // include { RUN_UTILITY                                       } from "../subworkflows/local/utility"
-include { VALIDATE_PARAMS                                   } from '../modules/local/general_util/validate_params/main'
+include { validateParameters; paramsSummaryLog; samplesheetToList } from 'plugin/nf-schema'
 
 include { GET_WAIT_TIME                                     } from "../modules/local/general_util/get_wait_time/main"
 
@@ -45,9 +45,12 @@ workflow TOSTADAS {
 		exit 0
 	}
 
-	// validate params
-	VALIDATE_PARAMS()
-
+	// validate input parameters
+    validateParameters()
+
+    // print summary of supplied parameters
+    log.info paramsSummaryLog(workflow)
+
 	// run metadata validation process
 	METADATA_VALIDATION ( 
 		params.meta_path