plot depths per group

bin/depth_group_comparison.py

@@ -0,0 +1,173 @@
+#!/usr/bin/env python
+
+"""
+Plot depth of all genes and specific genes per sample groups.
+
+This script plots the average depth at all consensus exons across all genes and for specific genes of interest, either in the panel or in a custom subset of genes, stratified by sample groups defined in the metadata.
+The output is stored in a pdf file.
+"""
+
+import click
+import pandas as pd
+from utils_plot import plots_general_config
+import matplotlib.pyplot as plt
+import matplotlib as mpl
+from matplotlib.backends.backend_pdf import PdfPages
+import seaborn as sns
+import ast
+import re
+
+mpl.rcParams.update({
+    'axes.titlesize': plots_general_config["title_fontsize"],
+    'axes.labelsize': plots_general_config["xylabel_fontsize"],
+    'xtick.labelsize': plots_general_config["xyticks_fontsize"],
+    'ytick.labelsize': plots_general_config["xyticks_fontsize"],
+    'figure.titlesize': plots_general_config["title_fontsize"],
+})
+
+separator2character = {
+    'tab' : '\t',
+    'comma' : ','
+}
+
+
+def plot_depth_per_group(df, group_col, data_type, pdf):
+    '''
+    Function to plot depth within a group of samples in all genes or a specific subset of genes
+    '''
+
+    col_name = group_col[0] if isinstance(group_col, list) else group_col
+
+    # Get the number of unique categories to plot
+    num_categories = df[col_name].nunique() + 2
+    plt.figure(figsize=(num_categories, 4))
+
+    if data_type == 'ALL_GENES':
+        plot_df = df[df['GENE'] == 'ALL_GENES']
+        title = f"Average Depth for ALL_GENES in {col_name} group"
+
+    else: # data_type is the Gene Name
+        plot_df = df[df['GENE'] == data_type]
+        title = f"Average Depth for {data_type} in {col_name} group"
+
+    if plot_df.empty:
+        print(f"No data available for {data_type} in {col_name} group. Skipping plot.")
+        return
+
+    sns.boxplot(data=plot_df, x=col_name, y="MEAN_GENE_DEPTH", hue=col_name, showfliers=False, showmeans=False, legend=False)
+    sns.stripplot(data=plot_df, x=col_name, y="MEAN_GENE_DEPTH", color='grey', alpha=0.5, size=4, legend=False)    
+
+    plt.title(title, fontsize=plots_general_config["title_fontsize"])                
+    plt.xlabel('', fontsize=plots_general_config["xylabel_fontsize"])
+    plt.ylabel(f"Average Cons Exons Depth", fontsize=plots_general_config["xylabel_fontsize"])
+    plt.yticks( fontsize=plots_general_config["yticks_fontsize"])
+    plt.xticks(fontsize=plots_general_config["xticks_fontsize"])
+    plt.tick_params(axis='x', rotation=90)
+    plt.tight_layout()
+    pdf.savefig()
+    plt.close()
+    plt.show()
+
+    return 
+
+
+@click.command()
+@click.option('--table-filename', required=True, type=click.Path(exists=True), help='Input features table file')
+@click.option('--depth-gene-sample', required=True, type=click.Path(exists=True), help='Input depth file per gene per sample')
+@click.option('--depth-sample', required=True, type=click.Path(exists=True), help='Input depth file per sample')
+@click.option('--separator', required=True, type=click.Choice(['tab', 'comma']), help='Separator used in features table: tab or comma')
+@click.option('--unique-identifier', default=None, type=str, help='Unique identifier column name')
+@click.option('--groups', required=True, type=str, help='List of columns with grouping information')
+@click.option('--custom-genes', required=False, type=str, help='Comma separated list of custom genes')
+
+
+def main(table_filename, depth_gene_sample, depth_sample, unique_identifier, separator, groups, custom_genes):
+
+    sep_char = separator2character[separator]
+    uniq_name = unique_identifier if unique_identifier else "sample"
+
+    # Read tables
+    features_table = pd.read_table(table_filename, header=0, sep=sep_char)
+    depth_genes_samples = pd.read_table(depth_gene_sample, header=0, sep="\t")
+    depth_per_sample = pd.read_table(depth_sample, header=0, sep="\t")
+
+    # Process panel genes 
+    panel_genes = sorted(set(depth_genes_samples['GENE'].unique()))
+    if custom_genes:
+        print(f'Custom genes provided, plotting custom genes only: {custom_genes}')
+        custom_gene_list = [g.strip() for g in custom_genes.split(",")]
+        panel_genes = sorted(set(custom_gene_list) & set(panel_genes))
+
+    else:
+        print(f'No custom genes provided, plotting all genes in the panel: {panel_genes}')
+
+
+    # Process depth per sample to add the 'ALL_GENES' depth value per sample
+    print('Processing per sample depth table to add the ALL_GENES depth column: ')
+    depth_per_sample['GENE'] = 'ALL_GENES'
+    depth_per_sample = depth_per_sample.rename(columns={'avg_depth_sample': 'MEAN_GENE_DEPTH'})
+
+    print('Depth per sample table after adding ALL_GENES value in GENES column:')
+    print(depth_per_sample.head())
+
+    depth_genes_samples = pd.concat([depth_genes_samples, depth_per_sample], ignore_index=True)
+    print('Added ALL_GENES depth values to depth genes samples table:')
+    print('Length of table:', len(depth_genes_samples))
+    print(depth_genes_samples.head())
+
+    # Merge data with metadata table to get the group information for each sample
+    merged_depth_df = pd.merge(features_table, depth_genes_samples, how='left', left_on=uniq_name, right_on='SAMPLE_ID')
+    print('Merged depth and metadata table:')
+    print(merged_depth_df.head())
+    print('Length of merged table:', len(merged_depth_df))
+
+    # Process groups
+    # First clean the string (adds quotes if the shell stripped them)
+    cleaned = re.sub(r'(?<!["\'])\b([a-zA-Z_][a-zA-Z0-9_]*)\b(?!["\'])', r'"\1"', groups)
+
+    # Then parse into a list of lists
+    raw_groups = ast.literal_eval(cleaned) if groups else []
+
+    # Finally flatten and get unique items
+    # This nested loop goes through each sub-list and each item within it
+    groups_of_interest = list(set(item for sublist in raw_groups for item in sublist))
+
+
+    print(f"Processing data for the groups of interest: {groups_of_interest}")
+
+    # Handle groups so each group has its own plot in all and individual genes and stored in the same pdf file per group
+    for group in groups_of_interest:
+        output_name = f"{group}.plot_depth_per_group.pdf"
+
+        with PdfPages(output_name) as pdf:    
+            print(f"Processing {group} group, type: {type(group)}")
+            group_table = merged_depth_df[[uniq_name, group, 'GENE', 'MEAN_GENE_DEPTH']]
+            print(group_table.head())
+            print('Length of the processed table', len(group_table))
+
+            # Plot depth of all samples for each group
+            plot_depth_per_group(group_table, group, 'ALL_GENES', pdf)
+
+            # Plot depths per gene (defined cutom genes or genes in the panel) for each group
+            gene_table = group_table[group_table['GENE'].isin(panel_genes)]
+            for gene in panel_genes:
+                gene_table = group_table[group_table['GENE'] == gene]
+                print('Length of gene data for gene', gene, ':', len(gene_table))
+                plot_depth_per_group(gene_table, group, gene, pdf)
+
+            print(f"Plots saved as {output_name}")
+
+if __name__ == "__main__":
+    main()
+
+'''
+Example usage:
+python depth_group_comparison.py \
+       --table-filename metadata_table_all_with_bacterial_signatures.tsv \
+        --depth-table all_samples.exons_cons.depth_per_gene_per_sample.tsv \
+        --depth-sample all_samples.exons_cons.depth_per_sample.tsv \
+        --separator tab \
+        --unique-identifier Sample_Name \
+        --groups "[ ["Sample_Group"], ["cancer"], ["Age_onset"], ["Cancer_age_group"] , ["Bacterial_Signatures_identified"]]" \
+        --custom-genes APC,BRAF,FBXW7,KRAS,PIK3CA,SMAD4,TP53' \
+'''

conf/modules.config

@@ -336,6 +336,21 @@ process {
         ]
     }
 
+    withName: "ANALYZEDEPTHSGROUPS" {
+        // these params.features are defined in nextflow.config
+        ext.unique_identifier           = params.features_unique_identifier
+        ext.features_groups             = params.features_groups_list
+        ext.separator                   = params.features_table_separator
+        ext.custom_genes                = params.features_genes_list
+        // define the list of custom genes here
+        // you will have to add an extra parameters to the pipeline (nextflow.config) and then handle it here
+        publishDir = [
+            path: { "${params.outdir}/plots/depths_summary/plots_per_group" }, 
+            mode: params.publish_dir_mode,
+            pattern: '**{pdf}',
+        ]
+    }
+
     withName: GROUPGENES {
         ext.custom    = params.custom_groups
         ext.hotspots  = params.create_subgenic_regions

modules/local/analyzedepths/main.nf

@@ -0,0 +1,54 @@
+process ANALYZE_DEPTHS_GROUPS {
+
+    tag "groups"
+    label 'process_low'
+
+    label 'deepcsa_core'
+
+    input:
+    path(features_table)
+    // note that these input depth files are generated in another step and defined in /deepCSA/subworkflows/local/plotdepths/main.nf
+    tuple val(meta) , path(average_depth_gene_sample) // needs to be added as a tupple since in PLOT_DEPTHS module (/modules/plot/depths_summary/main.nf) the output is set up as a tupple to"track" to which metadata belongs to this file
+    tuple val(meta) , path(average_depth_sample) 
+
+
+
+    output:
+    // the main outputs will be the PDFs
+    path("*.plot_depth_per_group.pdf")                           , emit: plots_per_gene_per_group
+
+    script:
+
+    def separator = task.ext.separator ?: "comma"
+    def custom_groups = task.ext.features_groups ? "--groups \"${task.ext.features_groups}\" " : ""
+    def custom_genes = task.ext.features_genes ? "--custom-genes \"${task.ext.features_genes}\" " : ""
+    def unique_identifier = task.ext.unique_identifier ? "--unique-identifier ${task.ext.unique_identifier}" : ""
+
+    // depth_group_comparison.py is in bin/ and has execution permissions add shebang 
+    // ${average_depth_gene_sample} comes from subworkflows/local/plotdepths/main.nf
+    """
+
+    depth_group_comparison.py \\
+                --table-filename ${features_table} \\
+                --depth-table ${average_depth_gene_sample} \\
+                --separator ${separator} \\
+                ${unique_identifier} \\
+                ${custom_groups} \\
+                ${custom_genes} \\
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        python: \$(python --version | sed 's/Python //g')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    touch depth_group_comparison.plot_depth_per_group.pdf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        python: \$(python --version | sed 's/Python //g')
+    END_VERSIONS
+    """
+}

nextflow.config

@@ -17,6 +17,7 @@ params {
     features_table_separator                 = 'comma'
     features_unique_identifier               = null
     features_groups_list                     = null
+    features_genes_list                      = null
 
     custom_groups                            = false
     custom_groups_file                       = null

workflows/deepcsa.nf

@@ -101,6 +101,8 @@ include { TABLE_2_GROUP                 as TABLE2GROUP              } from '../m
 include { ANNOTATE_DEPTHS               as ANNOTATEDEPTHS           } from '../modules/local/annotatedepth/main'
 include { DOWNSAMPLE_DEPTHS             as DOWNSAMPLEDEPTHS         } from '../modules/local/downsample/depths/main'
 
+include { ANALYZE_DEPTHS_GROUPS         as ANALYZEDEPTHSGROUPS      } from '../modules/local/analyzedepths/main'
+
 include { SELECT_MUTDENSITIES           as SYNMUTDENSITY            } from '../modules/local/select_mutdensity/main'
 include { SELECT_MUTDENSITIES           as SYNMUTREADSDENSITY       } from '../modules/local/select_mutdensity/main'
 
@@ -227,6 +229,11 @@ workflow DEEPCSA{
     }
     PLOTDEPTHSEXONSCONS(ANNOTATEDEPTHS.out.all_samples_depths, CREATEPANELS.out.exons_consensus_bed, CREATEPANELS.out.exons_consensus_panel)
 
+    // ANALYZEDEPTHSGROUPS should run only when user defines a group list
+    if (params.features_groups_list) {
+        ANALYZEDEPTHSGROUPS(features_table, PLOTDEPTHSEXONSCONS.out.average_depth_gene_sample)
+    }
+
     // Enrich regions in consensus panels
     ENRICHPANELS(MUT_PREPROCESSING.out.mutations_all_samples,
                  ANNOTATEDEPTHS.out.all_samples_depths,