eva-sub-cli 0.4.4__py3-none-any.whl

eva_sub_cli/jinja_templates/html_report.html

@@ -0,0 +1,142 @@
+<!DOCTYPE html>
+{% from 'project_details.html' import project_details_report %}
+{% from 'file_validation.html' import file_validation_report %}
+{% from 'sample_name_check.html' import sample_name_check_report %}
+{% from 'fasta_check.html' import fasta_check_report %}
+{% from 'metadata_validation.html' import metadata_validation_report %}
+{% from 'shallow_validation.html' import optional_shallow_validation_report %}
+
+<html lang="EN">
+<head>
+    <meta charset="UTF-8">
+    <title>Validation Report</title>
+    <style>
+        body { font-family: Verdana, sans-serif; }
+
+        .logo { float: left; padding: 20px; }
+        .title { padding: 8px; }
+        header { overflow: auto; }
+        header::after { clear: both; }
+
+        section { padding: 20px 20px 10px 40px; }
+        .description { padding-bottom: 10px}
+        .error-description { padding: 10px 0px 10px}
+
+        .report-section { padding: 12px; border: 1px solid white; margin-bottom: 5px}
+        .active, .collapsible:hover.fail { background-color: #e6a3ae; }
+        .active, .collapsible:hover.info { background-color: #c0c0c0; }
+
+        table { border-collapse: collapse; padding: 10px; margin-bottom: 20px}
+        th, td { border: 1px solid black;  padding: 8px;  text-align: left; }
+        th {  background-color: lightgrey;  }
+        .fail { background-color: #FFB6C1; }
+        .pass { background-color: #90EE90; }
+        .info { background-color: #dadada; }
+        .error-list, .no-show { display: none; }
+    </style>
+</head>
+<body>
+
+<header>
+    <div class="logo">
+    <img src="data:image/png;base64,{{ logo_data }}" width="100" height="100" alt="EVA logo">
+        </div>
+    <div class="title">
+        <h1>Validation Report</h1>
+        <h6>eva-sub-cli v{{cli_version}}</h6>
+    </div>
+</header>
+
+{{ optional_shallow_validation_report(validation_results) }}
+
+<section>
+    <h2>Project Summary</h2>
+    <div class="description">
+        General details about the project
+    </div>
+    {{ project_details_report(project_title, validation_date, submission_dir, vcf_fasta_analysis_mapping) }}
+</section>
+
+<section>
+    <h2>Metadata validation results</h2>
+    <div class="description">
+        Ensures that required fields are present and values are formatted correctly.
+        For requirements, please refer to the <a href="https://www.ebi.ac.uk/eva/?Submit-Data" target=”_blank”>EVA website</a>.
+    </div>
+    {{ metadata_validation_report(validation_results) }}
+</section>
+
+<section>
+    <h2>VCF validation results</h2>
+    <div class="description">
+        Checks whether each file is compliant with the <a href="http://samtools.github.io/hts-specs/VCFv4.3.pdf" target=”_blank”>VCF specification</a>.
+        Also checks whether the variants' reference alleles match against the reference assembly.
+    </div>
+    {% for file_name in vcf_files %}
+        {% if file_name!= "pass"%}
+        <h3>{{ file_name }}</h3>
+        {{ file_validation_report(validation_results, file_name) }}
+        {% endif %}
+    {% endfor %}
+</section>
+
+<section>
+    <h2>Sample name concordance check</h2>
+    <div class="description">
+        Checks whether information in the metadata is concordant with that contained in the VCF files, in particular sample names.
+    </div>
+    {{ sample_name_check_report(validation_results)}}
+</section>
+
+<section>
+    <h2>Reference genome INSDC check</h2>
+    <div class="description">
+        Checks that the reference sequences in the FASTA file used to call the variants are accessioned in INSDC.
+        Also checks if the reference assembly accession in the metadata matches the one determined from the FASTA file.
+    </div>
+    {% for file_name in fasta_files %}
+        {% if file_name!= "pass"%}
+        <h3>{{ file_name }}</h3>
+        {{ fasta_check_report(validation_results, file_name) }}
+        {% endif %}
+    {% endfor %}
+</section>
+
+<script>
+    let collapsibles = document.querySelectorAll('.collapsible');
+    for (let collapsible of collapsibles) {
+        collapsible.addEventListener('click', function() {
+            this.classList.toggle('active');
+
+             let icon = this.querySelector('.expand_icon');
+             if (this.classList.contains('active')) {
+                 icon.innerHTML = '&#9660;';
+             } else {
+                 icon.innerHTML = '&#9654;';
+             }
+
+            let content = this.nextElementSibling;
+            if (content.style.display === 'block') {
+                content.style.display = 'none';
+            }
+            else {
+                content.style.display = 'block';
+                content.scrollIntoView({block: 'nearest'});
+            }
+        });
+    }
+
+    function toggleSampleAllErrorsList(id) {
+        var element = document.getElementById(id);
+        var link = document.querySelector(`a[onclick="toggleSampleAllErrorsList('${id}'); return false;"]`);
+        if (element.style.display === "none" || element.style.display === "") {
+            element.style.display = "block";
+            link.innerHTML = "Hide All Errors for Category";
+        } else {
+            element.style.display = "none";
+            link.innerHTML = "Show All Errors for Category";
+        }
+    }
+</script>
+</body>
+</html>

eva_sub_cli/jinja_templates/metadata_validation.html

@@ -0,0 +1,52 @@
+
+{% macro metadata_validation_report(validation_results) -%}
+    {% set results = validation_results.get('metadata_check', {}) %}
+    {% set spreadsheet_errors = results.get('spreadsheet_errors', []) %}
+    {% set json_errors = results.get('json_errors', []) %}
+
+    {% set has_errors = spreadsheet_errors or json_errors %}
+    {% if has_errors %}
+        {% set expand_icon = "▶" %}
+        {% set icon = "❌" %}
+        {% set row_class = "report-section fail collapsible" %}
+    {% else %}
+        {% set icon = "✔" %}
+        {% set row_class = "report-section pass" %}
+    {% endif %}
+    <div class='{{ row_class }}'><span class="expand_icon">{{ expand_icon }}</span> {{ icon }} Metadata validation check </div>
+    {% if spreadsheet_errors %}
+        <div class="error-list">
+            <div class="error-description"><strong>Full report:</strong> {{ results.get('spreadsheet_report_path', '') }}</div>
+            <table>
+                <tr>
+                    <th>Sheet</th><th>Row</th><th>Column</th><th>Description</th>
+                </tr>
+                {% for error in spreadsheet_errors %}
+                    <tr>
+                        <td><strong>{{ error.get('sheet') }}</strong></td>
+                        <td><strong>{{ error.get('row') }}</strong></td>
+                        <td><strong>{{ error.get('column') }}</strong></td>
+                        <td> {{ error.get('description') }}</td>
+                    </tr>
+                {% endfor %}
+            </table>
+        </div>
+    {% endif %}
+
+    {% if json_errors %}
+        <div class="error-list">
+            <div class="error-description"><strong>Full report:</strong> {{ results.get('json_report_path', '') }}</div>
+            <table>
+                <tr>
+                    <th>JSON Property</th><th>Error Description</th>
+                </tr>
+                {% for error in json_errors %}
+                    <tr>
+                        <td><strong>{{ error.get('property') }}</strong></td>
+                        <td> {{ error.get('description') }}</td>
+                    </tr>
+                {% endfor %}
+            </table>
+        </div>
+    {% endif %}
+{%- endmacro %}

eva_sub_cli/jinja_templates/project_details.html

@@ -0,0 +1,37 @@
+{% macro project_details_report(project_title, validation_date, submission_dir, vcf_fasta_analysis_mapping) -%}
+    <div class="report-section info">
+          {% if project_title %}
+             <p><b>Project Title:</b> {{ project_title }}</p>
+         {% else %}
+             <p><b>Project Title:</b> Not Found</p>
+         {% endif %}
+        <p><b>Validation Date:</b> {{ validation_date }}</p>
+        <p><b>Submission Directory:</b> {{ submission_dir }}</p>
+    </div>
+
+    {% if vcf_fasta_analysis_mapping %}
+        {% set row_class = "report-section info collapsible" %}
+        {% set expand_icon = "&#9654;" %}
+        <div class='{{ row_class }}'><span class="expand_icon">{{ expand_icon }}</span> Files mapping</div>
+            <div class="no-show">
+                <table>
+                    <tr>
+                        <th><strong>VCF File</strong></th>
+                        <th><strong>Fasta File</strong></th>
+                        <th><strong>Analysis</strong></th>
+                    </tr>
+                    {% for mapping in vcf_fasta_analysis_mapping %}
+                    <tr>
+                        <td>{{ mapping.vcf_file }}</td>
+                        <td>{{ mapping.fasta_file }}</td>
+                        <td>{{ mapping.analysis }}</td>
+                    </tr>
+                    {% endfor %}
+                </table>
+            </div>
+        </div>
+    {% endif %}
+
+
+
+{%- endmacro %}

eva_sub_cli/jinja_templates/sample_name_check.html

@@ -0,0 +1,60 @@
+
+{% macro sample_name_check_report(validation_results) -%}
+    {% set results = validation_results.get('sample_check', {}) %}
+    {% macro format_error(error) %}
+        {{ error | replace(' ', '•') }}
+    {% endmacro %}
+
+    {% for analysis, results_for_analysis in results.get('results_per_analysis', {}).items() %}
+        {% if results_for_analysis.get('difference') %}
+            {% set expand_icon = "▶" %}
+            {% set icon = "❌" %}
+            {% set row_class = "report-section fail collapsible" %}
+            {% set text = "Sample names in metadata do not match with those in VCF files" %}
+        {% else %}
+            {% set icon = "✔" %}
+            {% set row_class = "report-section pass" %}
+            {% set text = "Sample names in metadata match with those in VCF files" %}
+        {% endif %}
+        <div class='{{ row_class }}'><span class="expand_icon">{{ expand_icon }}</span> {{ icon }} <b>{{ analysis }}</b>: {{ text }} </div>
+        {% if results_for_analysis.get('difference') %}
+            <div class="error-list">
+                <table>
+                    <tr>
+                        <th>Category</th><th>First 5 Errors For Category</th><th>Link To View All Errors</th>
+                    </tr>
+                    <tr>
+                        <td><strong>Samples described in the metadata but not in the VCF files</strong></td>
+                        <td>{{ results_for_analysis.get('more_metadata_submitted_files')[:5]|join(", ") }}</td>
+                        <td> <a href="#" onclick="toggleSampleAllErrorsList('metadata-{{ loop.index0 }}'); return false;">Show All Errors For Category</a></td>
+                    </tr>
+                    <tr>
+                        <td><strong>Samples in the VCF files but not described in the metadata</strong></td>
+                        <td>{{ results_for_analysis.get('more_submitted_files_metadata')[:5]|join(", ") }}</td>
+                        <td> <a href="#" onclick="toggleSampleAllErrorsList('vcf-{{ loop.index0 }}'); return false;">Show All Errors For Category</a></td>
+                    </tr>
+                </table>
+
+                <div id="metadata-{{ loop.index0 }}" class="description no-show">
+                    <div class="error-description"><strong>All Errors For Category - Samples described in the metadata but not in the VCF files:</strong></div>
+                    <ol>
+                        {% for error in results_for_analysis.get('more_metadata_submitted_files') %}
+                            <li>{{ format_error(error) }}</li>
+                        {% endfor %}
+                    </ol>
+                    <a href="#" onclick="toggleSampleAllErrorsList('metadata-{{ loop.index0 }}'); return false;">Hide</a>
+                </div>
+
+                <div id="vcf-{{ loop.index0 }}" class="description no-show">
+                    <div class="error-description"><strong>All Errors For Category - Samples in the VCF files but not described in the metadata:</strong></div>
+                    <ol>
+                        {% for error in results_for_analysis.get('more_submitted_files_metadata') %}
+                            <li>{{ format_error(error) }}</li>
+                        {% endfor %}
+                    </ol>
+                    <a href="#" onclick="toggleSampleAllErrorsList('vcf-{{ loop.index0 }}'); return false;">Hide</a>
+                </div>
+            </div>
+        {% endif %}
+    {% endfor %}
+{%- endmacro %}

eva_sub_cli/jinja_templates/shallow_validation.html

@@ -0,0 +1,29 @@
+
+{% macro optional_shallow_validation_report(validation_results) -%}
+    {% set results = validation_results.get('shallow_validation', {}) %}
+
+    {% if results.get('required') %}
+    <section>
+        <div class="report-section fail collapsible"> <span class="expand_icon">&#9654;</span>
+            &#10060; <b>You requested to run the shallow validation, please run full validation before submitting the data</b>
+        </div>
+        <div class="no-show">
+        <table>
+            <tr>
+                <th><strong>VCF File</strong></th>
+                <th><strong>Variant lines validated in VCF</strong></th>
+                <th><strong>Entries used in Fasta</strong></th>
+            </tr>
+            {% for vcf_file in results.get('metrics') %}
+                <tr>
+                    <td>{{ vcf_file }}</td>
+                    <td>{{ results.get('metrics').get(vcf_file).get('trim_down_vcf_record') }}</td>
+                    <td>{{ results.get('metrics').get(vcf_file).get('number_sequence_found') }}</td>
+                </tr>
+                {% endfor %}
+            </table>
+        </div>
+    </section>
+    {% endif %}
+
+{%- endmacro %}

eva_sub_cli/metadata_utils.py

@@ -0,0 +1,46 @@
+import os
+from collections import defaultdict
+
+from ebi_eva_common_pyutils.logger import logging_config
+
+logger = logging_config.get_logger(__name__)
+
+
+def get_samples_per_analysis(metadata):
+    """Returns mapping of analysis alias to sample names, based on metadata."""
+    samples_per_analysis = defaultdict(list)
+    for sample_info in metadata.get('sample', []):
+        for analysis_alias in sample_info.get('analysisAlias', []):
+            samples_per_analysis[analysis_alias].append(sample_info.get('sampleInVCF'))
+    return {
+        analysis_alias: set(samples)
+        for analysis_alias, samples in samples_per_analysis.items()
+    }
+
+
+def get_files_per_analysis(metadata):
+    """Returns mapping of analysis alias to filenames, based on metadata."""
+    files_per_analysis = defaultdict(list)
+    for file_info in metadata.get('files', []):
+        files_per_analysis[file_info.get('analysisAlias')].append(file_info.get('fileName'))
+    return {
+        analysis_alias: set(filepaths)
+        for analysis_alias, filepaths in files_per_analysis.items()
+    }
+
+
+def get_reference_assembly_for_analysis(metadata, analysis_alias):
+    """Returns the reference assembly for this analysis (does not validate format)."""
+    for analysis in metadata.get('analysis', []):
+        if analysis.get('analysisAlias') == analysis_alias:
+            return analysis.get('referenceGenome')
+    return None
+
+
+def get_analysis_for_vcf_file(vcf_file, files_per_analysis):
+    """Returns list of analysis aliases associated with the vcf file path."""
+    if not os.path.exists(vcf_file):
+        raise FileNotFoundError(f'{vcf_file} cannot be resolved')
+    analysis_aliases = [analysis_alias for analysis_alias in files_per_analysis
+                        if os.path.basename(vcf_file) in files_per_analysis[analysis_alias]]
+    return analysis_aliases

eva_sub_cli/nextflow/validation.nf

@@ -0,0 +1,303 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl=2
+
+def helpMessage() {
+    log.info"""
+    Validate a set of VCF files and metadata to check if they are valid to be submitted to EVA.
+
+    Inputs:
+            --vcf_files_mapping     csv file with the mappings for vcf files, fasta and assembly report
+            --output_dir            output_directory where the reports will be output
+            --metadata_json         Json file describing the project, analysis, samples and files
+            --metadata_xlsx         Excel file describing the project, analysis, samples and files
+    """
+}
+
+params.vcf_files_mapping = null
+params.output_dir = null
+params.metadata_json = null
+params.metadata_xlsx = null
+
+// executables - external tools
+params.executable = [
+    "vcf_validator": "vcf_validator",
+    "vcf_assembly_checker": "vcf_assembly_checker",
+    "biovalidator": "biovalidator"
+]
+// python scripts - installed as part of eva-sub-cli
+params.python_scripts = [
+    "samples_checker": "samples_checker.py",
+    "fasta_checker": "check_fasta_insdc.py",
+    "xlsx2json": "xlsx2json.py",
+    "semantic_checker": "check_metadata_semantics.py",
+    "trim_down": "trim_down.py"
+]
+// prefix to prepend to all provided path
+params.base_dir = ""
+// help
+params.help = null
+params.shallow_validation = false
+
+// Show help message
+if (params.help) exit 0, helpMessage()
+
+// Test input files
+if (!params.vcf_files_mapping || !params.output_dir || (!params.metadata_json && !params.metadata_xlsx)) {
+    if (!params.vcf_files_mapping)      log.warn('Provide a csv file with the mappings (vcf, fasta, assembly report) --vcf_files_mapping')
+    if (!params.metadata_xlsx)           log.warn('Provide a json file with the metadata description of the project and analysis --metadata_json')
+    if (!params.output_dir)             log.warn('Provide an output directory where the reports will be copied using --output_dir')
+    if (!params.metadata_json && !params.metadata_xlsx)   log.warn('Provide a json or Excel file with the metadata description of the project and analysis --metadata_json or --metadata_xlsx')
+    exit 1, helpMessage()
+}
+
+schema_dir = file(projectDir).parent + '/etc'
+conversion_configuration = schema_dir + '/spreadsheet2json_conf.yaml'
+
+def joinBasePath(path) {
+    if (path){
+        return params.base_dir + '/' + path
+    }
+    return 'NO_FILE'
+}
+
+output_dir = joinBasePath(params.output_dir)
+
+workflow {
+    // Prepare the file path
+    vcf_and_ref_ch = Channel.fromPath(joinBasePath(params.vcf_files_mapping))
+        .splitCsv(header:true)
+        .map{row -> tuple(
+            file(joinBasePath(row.vcf)),
+            file(joinBasePath(row.fasta)),
+            file(joinBasePath(row.report))
+        )}
+    if (params.shallow_validation){
+        // create a smaller vcf and fasta then replace the channel
+        trim_down_vcf(vcf_and_ref_ch)
+        vcf_and_ref_ch = trim_down_vcf.out.vcf_and_ref
+    }
+    vcf_files = vcf_and_ref_ch.map{row -> row[0]}
+    fasta_to_vcfs = vcf_and_ref_ch.map{row -> tuple(row[1], row[0])}.groupTuple(by:0)
+    // VCF checks
+    check_vcf_valid(vcf_and_ref_ch)
+    check_vcf_reference(vcf_and_ref_ch)
+
+    generate_file_size_and_md5_digests(vcf_files)
+    collect_file_size_and_md5(generate_file_size_and_md5_digests.out.file_size_and_digest_info.collect())
+
+
+    // Metadata conversion
+    if (params.metadata_xlsx && !params.metadata_json){
+        convert_xlsx_2_json(joinBasePath(params.metadata_xlsx))
+        metadata_json = convert_xlsx_2_json.out.metadata_json
+    } else {
+        metadata_json = joinBasePath(params.metadata_json)
+    }
+    if (metadata_json) {
+        // Metadata checks and concordance checks
+        metadata_json_validation(metadata_json)
+        metadata_semantic_check(metadata_json)
+        sample_name_concordance(metadata_json, vcf_files.collect())
+        insdc_checker(metadata_json, fasta_to_vcfs)
+    }
+}
+
+
+process trim_down_vcf {
+    publishDir output_dir, overwrite: false, mode: "copy", pattern: "*.log"
+    publishDir output_dir, overwrite: false, mode: "copy", pattern: "*.yml"
+
+    input:
+    tuple path(vcf), path(fasta), path(report)
+
+    output:
+    tuple path("output/$vcf"), path("output/$fasta"), path(report), emit: vcf_and_ref
+    path "${vcf.getBaseName()}_trim_down.log", emit: trim_down_log
+    path "${vcf.getBaseName()}_trim_down.yml", emit: trim_down_metric
+
+    """
+    mkdir output
+    $params.python_scripts.trim_down --vcf_file $vcf  --output_vcf_file output/$vcf --fasta_file $fasta --output_fasta_file output/$fasta --output_yaml_file ${vcf.getBaseName()}_trim_down.yml > ${vcf.getBaseName()}_trim_down.log
+    # This is needed to ensure that a missing (NO_FILE) report can still be passed down to subsequent steps
+    touch $report
+    """
+
+}
+
+/*
+* Validate the VCF file format
+*/
+process check_vcf_valid {
+    publishDir output_dir,
+            overwrite: false,
+            mode: "copy"
+
+    input:
+    tuple path(vcf), path(fasta), path(report)
+
+    output:
+    path "vcf_format/*.errors.*.txt", emit: vcf_validation_txt
+    path "vcf_format/*.vcf_format.log", emit: vcf_validation_log
+
+    """
+    trap 'if [[ \$? == 1 ]]; then exit 0; fi' EXIT
+
+    mkdir -p vcf_format
+    $params.executable.vcf_validator -i $vcf -r text -o vcf_format --require-evidence > vcf_format/${vcf}.vcf_format.log 2>&1
+    """
+}
+
+/*
+* Validate the VCF reference allele
+*/
+process check_vcf_reference {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    tuple path(vcf), path(fasta), path(report)
+
+    output:
+    path "assembly_check/*text_assembly_report*", emit: assembly_check_report
+    path "assembly_check/*.assembly_check.log", emit: assembly_check_log
+
+    script:
+    def report_opt = report.name != 'NO_FILE' ? "-a $report" : ''
+
+    """
+    trap 'if [[ \$? == 1 || \$? == 139 ]]; then exit 0; fi' EXIT
+
+    mkdir -p assembly_check
+    $params.executable.vcf_assembly_checker -i $vcf -f $fasta $report_opt -r summary,text  -o assembly_check --require-genbank > assembly_check/${vcf}.assembly_check.log 2>&1
+    """
+}
+
+process generate_file_size_and_md5_digests {
+    input:
+    path(vcf_file)
+
+    output:
+    path "${vcf_file}_file_size_and_digest_info.txt", emit: file_size_and_digest_info
+
+    script:
+    // Capture the realpath of the vcf to be able to resolve the file based on path instead of name
+    """
+     echo "\$(md5sum $vcf_file | awk '{print \$1}') \$(stat -c%s \$(readlink -f $vcf_file)) \$(readlink -f $vcf_file)" > ${vcf_file}_file_size_and_digest_info.txt
+    """
+}
+
+process collect_file_size_and_md5 {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    path(file_size_and_digests)
+
+    output:
+    path "file_info.txt", emit: file_info_log
+
+    script:
+    """
+    cat $file_size_and_digests > file_info.txt
+    """
+}
+
+
+process convert_xlsx_2_json {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    path(metadata_xlsx)
+
+    output:
+    path "metadata.json", emit: metadata_json, optional: true
+    path "metadata_conversion_errors.yml", emit: errors_yaml
+    path "xlsx2json.log", emit: xlsx2json_log
+
+    script:
+    metadata_json = metadata_xlsx.getBaseName() + '.json'
+
+    """
+    $params.python_scripts.xlsx2json --metadata_xlsx $metadata_xlsx --metadata_json metadata.json --errors_yaml metadata_conversion_errors.yml --conversion_configuration $conversion_configuration > xlsx2json.log 2>&1
+    """
+}
+
+process metadata_json_validation {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    path(metadata_json)
+
+    output:
+    path "metadata_validation.txt", emit: metadata_validation
+
+    script:
+    """
+    $params.executable.biovalidator --schema $schema_dir/eva_schema.json --ref $schema_dir/eva-biosamples.json --data $metadata_json > metadata_validation.txt 2>&1
+    """
+}
+
+process sample_name_concordance {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    path(metadata_json)
+    path(vcf_files)
+
+    output:
+    path "sample_checker.yml", emit: sample_checker_yml
+    path "sample_checker.log", emit: sample_checker_log
+
+    script:
+    """
+    $params.python_scripts.samples_checker --metadata_json $metadata_json --vcf_files $vcf_files --output_yaml sample_checker.yml > sample_checker.log 2>&1
+    """
+}
+
+
+process insdc_checker {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    path(metadata_json)
+    tuple(path(fasta_file), path(vcf_files))
+
+    output:
+    path "${fasta_file}_check.yml", emit: fasta_checker_yml
+    path "fasta_checker.log", emit: fasta_checker_log
+
+    script:
+    """
+    $params.python_scripts.fasta_checker --metadata_json $metadata_json --vcf_files $vcf_files --input_fasta $fasta_file --output_yaml ${fasta_file}_check.yml > fasta_checker.log 2>&1
+    """
+}
+
+
+process metadata_semantic_check {
+    publishDir output_dir,
+            overwrite: true,
+            mode: "copy"
+
+    input:
+    path(metadata_json)
+
+    output:
+    path "metadata_semantic_check.yml", emit: metadata_semantic_check_yml
+    path "semantic_checker.log", emit: semantic_checker_log
+
+    script:
+    """
+    $params.python_scripts.semantic_checker --metadata_json $metadata_json --output_yaml metadata_semantic_check.yml > semantic_checker.log 2>&1
+    """
+}