bam-filtration

Performs data cleaning and removal operations. This skill takes a raw BAM and creates a new, "clean" BAM file by actively removing artifacts: mitochondrial reads, blacklisted regions, PCR duplicates, and unmapped reads. Use this skill to "clean," "filter," or "remove bad reads" from a dataset. This is a prerequisite step before peak calling. Do NOT use this skill if you only want to view statistics without modifying the file.

⭐2stars🔀1forks📁View on GitHub🕐Updated Dec 12, 2025⬇️1downloads

When & Why to Use This Skill

This Claude skill automates the essential preprocessing of genomic data by cleaning and filtering BAM files. Specifically designed for ChIP-seq and ATAC-seq workflows, it removes technical artifacts such as mitochondrial reads, PCR duplicates, and blacklisted regions to ensure high-quality inputs for peak calling and downstream analysis.

Use Cases

  • Pre-processing ChIP-seq or ATAC-seq datasets to remove noise and technical artifacts before performing peak calling.
  • Automating the removal of mitochondrial DNA and unmapped reads to improve the accuracy of genomic research results.
  • Filtering out blacklisted genomic regions to prevent false-positive signals in high-throughput sequencing analysis.
  • Streamlining bioinformatics pipelines by transforming raw, coordinate-sorted BAM files into 'clean' datasets ready for publication-quality analysis.
nameBAM-filtration
descriptionPerforms data cleaning and removal operations. This skill takes a raw BAM and creates a new, "clean" BAM file by actively removing artifacts: mitochondrial reads, blacklisted regions, PCR duplicates, and unmapped reads. Use this skill to "clean," "filter," or "remove bad reads" from a dataset. This is a prerequisite step before peak calling. Do NOT use this skill if you only want to view statistics without modifying the file.

BAM Filtration for ChIP-seq / ATAC-seq

Overview

Main steps include:

  • Check the availability of blacklist file in current directory. Always prompt user whether to filter blacklist if blacklist files are missing. if the user need to filter blacklist file, then prompt user for the path of blacklist file.
  • Initialize the project directory and create the required directory.
  • Refer to the Inputs & Outputs section to check inputs and build the output architecture. All the output file should located in ${proj_dir} in Step 0.
  • Discover input BAMs in the current directory (or those matching a target token), and only select BAMs that are already coordinate-sorted and contain read group (RG) information.
  • Perform the filtration task with tools.

When to use this skill

  • Use this skill to "clean," "filter," or "remove bad reads" from a dataset
  • This is a prerequisite step before peak calling.
  • Do NOT use this skill if you only want to view statistics without modifying the file.

Inputs & Outputs

Inputs

${sample}.bam # BAMs that are already coordinate-sorted and contain read group (RG) information

Outputs

all_bam_filtration/
  filtered_bam/
    ${sample}.filtered.bam
    ${sample}.filtered.bam.bai
  temp/
    ... # intermediate files

Decision Tree

Step 0: Initialize Project

Call:

  • mcp__project-init-tools__project_init

with:

  • sample: all
  • task: bam_filtration

The tool will:

  • Create ${sample}_bam_filtration directory.
  • Return the full path of the ${sample}_bam_filtration directory, which will be used as ${proj_dir}.

Step 1: Filter BAM files

Call:

  • mcp__qc-tools__bam_artifacts

with:

  • bam_file: BAMs that are already coordinate-sorted and contain read group (RG) information
  • output_bam: ${proj_dir}/filtered_bam/${sample}.filtered.bam
  • temp_dir: ${proj_dir}/temp/
  • blacklist_bed: Path of the blacklist file