Step-by-Step Guide: Resolving “Out of Disk Space” Issues with Picard Tools

Running bioinformatics tools like Picard can sometimes lead to “out of disk space” errors, even when you have ample storage. This guide will walk you through the steps to resolve such issues, including tips, tricks, and scripts to manage disk space effectively.

1. Identify the Problem

The error java.io.IOException: No space left on device typically occurs when the temporary directory used by Java (or Picard) is on a partition with insufficient space. This is common in high-performance computing environments where /tmp is small.

2. Check Disk Space

Before proceeding, verify the available disk space on your system.

Unix Command:

df -h

This will display the disk usage and available space on all mounted partitions. Look for the partition where /tmp is located (usually /).

3. Create a Custom Temporary Directory

To avoid using the default /tmp directory, create a custom temporary directory in a location with sufficient space.

Unix Command:

mkdir -p /path/to/large/disk/tmp

Replace /path/to/large/disk/tmp with a directory path on a partition with ample space.

4. Run Picard Tools with a Custom Temporary Directory

Use the -Djava.io.tmpdir Java option to specify the custom temporary directory. Additionally, Picard Tools often supports a TMP_DIR parameter.

Example Command:

java -Xmx2g -Djava.io.tmpdir=/path/to/large/disk/tmp -jar SortSam.jar \
  SORT_ORDER=coordinate \
  INPUT=input.bam \
  OUTPUT=output.sorted.bam \
  TMP_DIR=/path/to/large/disk/tmp

Explanation:

• -Xmx2g: Allocates 2GB of memory to the Java process.
• -Djava.io.tmpdir=/path/to/large/disk/tmp: Sets the Java temporary directory.
• TMP_DIR=/path/to/large/disk/tmp: Ensures Picard uses the specified directory for temporary files.

5. Handle “Too Many Open Files” Error

If you encounter the error Too many open files, it means the system has hit the limit for open file handles. Increase the limit temporarily or permanently.

Temporary Fix (for current session):

ulimit -n 65536

Permanent Fix (requires root access):

Edit /etc/security/limits.conf and add the following lines:

* hard nofile 65536
* soft nofile 65536

Then, restart your session or reboot the system.

6. Using GATK with Picard Tools

If you’re running Picard Tools via GATK, the syntax for specifying the temporary directory is slightly different.

Example Command:

./gatk --java-options "-Djava.io.tmpdir=/path/to/large/disk/tmp" SortSam \
  -I=input.sam \
  -O=output.sorted.bam \
  -SO=coordinate \
  --TMP_DIR=/path/to/large/disk/tmp

Explanation:

• --java-options: Passes Java-specific options.
• --TMP_DIR: Specifies the temporary directory for GATK/Picard.

7. Automate with Scripts

To streamline the process, you can create a script to handle temporary directory creation and tool execution.

Bash Script:

#!/bin/bash

# Set variables
INPUT_BAM="input.bam"
OUTPUT_BAM="output.sorted.bam"
TMP_DIR="/path/to/large/disk/tmp"

# Create temporary directory
mkdir -p $TMP_DIR

# Run Picard SortSam
java -Xmx2g -Djava.io.tmpdir=$TMP_DIR -jar SortSam.jar \
  SORT_ORDER=coordinate \
  INPUT=$INPUT_BAM \
  OUTPUT=$OUTPUT_BAM \
  TMP_DIR=$TMP_DIR

Python Script:

import subprocess
import os

# Set variables
input_bam = "input.bam"
output_bam = "output.sorted.bam"
tmp_dir = "/path/to/large/disk/tmp"

# Create temporary directory
os.makedirs(tmp_dir, exist_ok=True)

# Run Picard SortSam
command = [
    "java", "-Xmx2g", f"-Djava.io.tmpdir={tmp_dir}", "-jar", "SortSam.jar",
    "SORT_ORDER=coordinate", f"INPUT={input_bam}", f"OUTPUT={output_bam}",
    f"TMP_DIR={tmp_dir}"
]

subprocess.run(command)

8. Monitor Disk Usage

Regularly monitor disk usage to avoid future issues.

Unix Command:

du -sh /path/to/large/disk/tmp

This will show the total size of the temporary directory.

9. Clean Up Temporary Files

After the job completes, clean up the temporary directory to free up space.

Unix Command:

rm -rf /path/to/large/disk/tmp/*

10. Additional Tips

• Compress Intermediate Files: Use tools like gzip or bzip2 to compress intermediate files if disk space is a concern.
• Use Cloud Storage: If working in a cloud environment, consider using high-capacity storage buckets for temporary files.
• Optimize Memory Usage: Adjust -Xmx (maximum memory) based on your system’s RAM to avoid excessive disk swapping.

By following these steps, you can effectively manage disk space issues when running Picard Tools and ensure smooth execution of your bioinformatics workflows.