CNV Annotation Usage Guide

CNV Annotation Usage Guide

Overview

CNV annotation adds biological context to copy number calls by identifying affected genes, pathways, and clinical significance. This is essential for interpreting CNV findings in research and clinical contexts.

Prerequisites

conda install -c bioconda annotsv pybedtools
pip install pandas

Quick Start

Tell your AI agent what you want to do:

• "Annotate my CNV segments with overlapping genes and cancer gene flags"
• "Filter out common CNVs using population frequency data from DGV"
• "Run AnnotSV on my structural variant VCF for comprehensive annotation"
• "Add ClinVar pathogenicity annotations to my CNV calls"

Annotation Resources

Complete Annotation Workflow

import pandas as pd
import pybedtools
from pathlib import Path

class CNVAnnotator:
    def __init__(self, gene_gtf, cgc_file=None, clinvar_vcf=None):
        self.genes = self.load_genes(gene_gtf)
        self.cancer_genes = self.load_cgc(cgc_file) if cgc_file else set()
        self.clinvar = clinvar_vcf

    def load_genes(self, gtf_file):
        '''Load gene annotations from GTF.'''
        genes = []
        with open(gtf_file) as f:
            for line in f:
                if line.startswith('#'):
                    continue
                fields = line.strip().split('\t')
                if fields[2] == 'gene':
                    attrs = dict(x.strip().split(' ', 1) for x in fields[8].split(';') if x.strip())
                    gene_name = attrs.get('gene_name', '').strip('"')
                    genes.append({
                        'chrom': fields[0],
                        'start': int(fields[3]),
                        'end': int(fields[4]),
                        'gene_name': gene_name
                    })
        return pd.DataFrame(genes)

    def load_cgc(self, cgc_file):
        '''Load COSMIC Cancer Gene Census.'''
        cgc = pd.read_csv(cgc_file, sep='\t')
        return set(cgc['Gene Symbol'].dropna().tolist())

    def annotate(self, cns_file):
        '''Annotate CNV segments.'''
        cns = pd.read_csv(cns_file, sep='\t')

        # BedTools intersection
        cns_bed = pybedtools.BedTool.from_dataframe(
            cns[['chromosome', 'start', 'end', 'log2']])
        genes_bed = pybedtools.BedTool.from_dataframe(
            self.genes[['chrom', 'start', 'end', 'gene_name']])

        intersect = cns_bed.intersect(genes_bed, wa=True, wb=True)

        # Parse results
        results = []
        for interval in intersect:
            results.append({
                'chrom': interval[0],
                'start': int(interval[1]),
                'end': int(interval[2]),
                'log2': float(interval[3]),
                'gene_name': interval[7]
            })

        df = pd.DataFrame(results)

        # Aggregate
        annotated = df.groupby(['chrom', 'start', 'end', 'log2']).agg({
            'gene_name': lambda x: ','.join(sorted(set(x)))
        }).reset_index()

        # Add cancer gene flags
        annotated['is_cancer_gene'] = annotated['gene_name'].apply(
            lambda x: any(g in self.cancer_genes for g in x.split(',')))

        annotated['cancer_genes'] = annotated['gene_name'].apply(
            lambda x: ','.join(g for g in x.split(',') if g in self.cancer_genes))

        # Add CNV type
        annotated['cnv_type'] = annotated['log2'].apply(self.classify_cnv)

        return annotated

    @staticmethod
    def classify_cnv(log2):
        if log2 < -1.0:
            return 'deep_deletion'
        elif log2 < -0.3:
            return 'deletion'
        elif log2 > 0.8:
            return 'high_amplification'
        elif log2 > 0.3:
            return 'amplification'
        return 'neutral'

# Usage
annotator = CNVAnnotator(
    gene_gtf='genes.gtf',
    cgc_file='cancer_gene_census.tsv'
)
annotated = annotator.annotate('sample.cns')
annotated.to_csv('sample_annotated.tsv', sep='\t', index=False)

Using AnnotSV

AnnotSV provides comprehensive annotation with a single command:

# Install
conda install -c bioconda annotsv

# Download annotations (first time)
AnnotSV -SVinputFile dummy.vcf -genomeBuild GRCh38

# Annotate CNVs
AnnotSV \
    -SVinputFile sample.cnv.vcf \
    -genomeBuild GRCh38 \
    -outputFile sample_annotsv \
    -SVminSize 1000 \
    -annotationMode full

# Key output columns:
# - Gene_name: overlapping genes
# - DGV_GAIN/LOSS_Freq: population frequency
# - ClinVar_clndn: disease associations
# - ACMG_class: pathogenicity classification

Cancer-Specific Annotation

def annotate_for_cancer(cnv_annotated, cancer_type='breast'):
    '''Add cancer-specific annotations.'''
    # Known driver genes by cancer type
    cancer_drivers = {
        'breast': ['BRCA1', 'BRCA2', 'TP53', 'ERBB2', 'PIK3CA', 'MYC'],
        'lung': ['EGFR', 'KRAS', 'ALK', 'MET', 'ROS1', 'TP53'],
        'colon': ['APC', 'KRAS', 'TP53', 'SMAD4', 'PIK3CA'],
    }

    drivers = set(cancer_drivers.get(cancer_type, []))

    cnv_annotated['driver_genes'] = cnv_annotated['gene_name'].apply(
        lambda x: ','.join(g for g in x.split(',') if g in drivers))

    cnv_annotated['has_driver'] = cnv_annotated['driver_genes'].str.len() > 0

    # Prioritize
    cnv_annotated['priority'] = cnv_annotated.apply(
        lambda x: 'high' if x['has_driver'] or abs(x['log2']) > 1
            else ('medium' if x['is_cancer_gene'] else 'low'), axis=1)

    return cnv_annotated

Population Frequency Filtering

def filter_by_frequency(cnv_annotated, dgv_file, max_freq=0.01):
    '''Filter common CNVs using DGV data.'''
    dgv = pybedtools.BedTool(dgv_file)

    cnv_bed = pybedtools.BedTool.from_dataframe(
        cnv_annotated[['chrom', 'start', 'end']])

    # Find CNVs overlapping common variants (50% reciprocal overlap)
    common = cnv_bed.intersect(dgv, f=0.5, r=True, u=True)
    common_coords = set([(str(i[0]), int(i[1]), int(i[2])) for i in common])

    cnv_annotated['is_common'] = cnv_annotated.apply(
        lambda x: (x['chrom'], x['start'], x['end']) in common_coords, axis=1)

    # Filter
    rare_cnvs = cnv_annotated[~cnv_annotated['is_common']]

    return rare_cnvs

Generate Summary Report

def generate_report(annotated, output_prefix):
    '''Generate annotation summary report.'''
    # Statistics
    stats = {
        'Total CNVs': len(annotated),
        'Amplifications': (annotated['cnv_type'].str.contains('amplification')).sum(),
        'Deletions': (annotated['cnv_type'].str.contains('deletion')).sum(),
        'Affecting cancer genes': annotated['is_cancer_gene'].sum(),
        'Total genes affected': annotated['gene_name'].str.split(',').explode().nunique(),
    }

    with open(f'{output_prefix}_summary.txt', 'w') as f:
        for k, v in stats.items():
            f.write(f'{k}: {v}\n')

    # High-priority CNVs
    high_priority = annotated[annotated['priority'] == 'high']
    high_priority.to_csv(f'{output_prefix}_high_priority.tsv', sep='\t', index=False)

    # Gene list for pathway analysis
    genes = annotated['gene_name'].str.split(',').explode().unique()
    with open(f'{output_prefix}_genes.txt', 'w') as f:
        f.write('\n'.join(sorted(genes)))

    return stats

Example Prompts

"Annotate my CNV segments with overlapping genes and cancer gene census flags"

"Filter out common CNVs from my callset using DGV population data"

"Add ClinVar annotations to my CNV calls and prioritize pathogenic variants"

"Run AnnotSV on my structural variant VCF to get comprehensive annotations"

Integration with Pathway Analysis

library(clusterProfiler)

# Load annotated CNV genes
amp_genes <- read.delim('cnv_annotated.tsv')
amp_genes <- amp_genes[amp_genes$log2 > 0.5, ]
genes <- unique(unlist(strsplit(amp_genes$gene_name, ',')))

# Convert to Entrez
entrez <- bitr(genes, fromType='SYMBOL', toType='ENTREZID', OrgDb='org.Hs.eg.db')

# Pathway enrichment
kegg <- enrichKEGG(gene=entrez$ENTREZID, organism='hsa')
go <- enrichGO(gene=entrez$ENTREZID, OrgDb='org.Hs.eg.db',

# Visualize
dotplot(kegg, showCategory=15)
dotplot(go, showCategory=15)