HRD Prediction Using CNVkit in Tumor-Only Mode: Step-by-Step Guide

can you describe steps for HRD prediction with using CNVkit in tumor only mode and suggest any software for this

Certainly! HRD (Homologous Recombination Deficiency) prediction using CNVkit in tumor-only mode involves several steps. CNVkit is a tool designed for copy number variation (CNV) analysis from high-throughput sequencing data. Here’s a step-by-step guide to using CNVkit for HRD prediction in tumor-only mode:

Step-by-Step Guide:

1. Install CNVkit:

   • Ensure you have Python installed.
   • Install CNVkit using pip:

     pip install cnvkit
     

2. Prepare Reference:

   • Since you are working in tumor-only mode, you need a reference genome. CNVkit can create a flat reference if you don't have matched normal samples.

     cnvkit.py reference -f <reference_genome.fasta> -o flat_reference.cnn
     

3. Run CNVkit on Tumor Sample:

   • Use CNVkit to process your tumor sample BAM file.

     cnvkit.py batch <tumor_sample.bam> -r flat_reference.cnn -d output_directory/
     

4. Segment the Copy Number Data:

   • Segment the copy number data to identify regions of gain or loss.

     cnvkit.py segment output_directory/<tumor_sample.cnr> -o output_directory/<tumor_sample.cns>
     

5. Call Copy Number Variants:

   • Call CNVs from the segmented data.

     cnvkit.py call output_directory/<tumor_sample.cns> -o output_directory/<tumor_sample.call.cns>
     

6. Analyze HRD:

   • For HRD prediction, you need to analyze the patterns of CNVs. Tools like scarHRD can be used to score HRD based on CNV data.
   • Convert CNVkit output to a format compatible with scarHRD or similar tools.

     cnvkit.py export seg output_directory/<tumor_sample.call.cns> -o output_directory/<tumor_sample.seg>
     

7. Run HRD Prediction Tool:

   • Use an HRD prediction tool like scarHRD to analyze the CNV data.

     library(scarHRD)
     scar_score <- scarHRD::scar_score("output_directory/<tumor_sample.seg>", reference_genome = "hg19")
     print(scar_score)
     

Suggested Software for HRD Prediction:

1. scarHRD:

   • An R package that calculates HRD scores based on CNV data.
   • Installation:

     install.packages("scarHRD")
     

2. HRDetect:

   • A more comprehensive tool that integrates multiple types of genomic data (including CNVs) to predict HRD.
   • Requires more complex setup and integration with other data types.

3. LOH-HRD:

   • Another R package that focuses on loss of heterozygosity (LOH) as a marker for HRD.
   • Installation:

     install.packages("LOH-HRD")
     

Summary:

• Install CNVkit and prepare a flat reference.
• Process the tumor sample to identify CNVs.
• Segment and call CNVs using CNVkit.
• Export CNV data to a compatible format for HRD prediction tools.
• Use HRD prediction tools like scarHRD to analyze the CNV data and predict HRD status.

By following these steps, you can effectively use CNVkit in tumor-only mode to predict HRD and utilize additional software to interpret the results.