SAMtools count

View full source code on GitHub

This applet performs a basic samtools view -c {bam} command, referred to as “SAMtools count”, on the DNAnexus platform.

Download BAM Files

For bash scripts, inputs to a job execution become environment variables. The inputs from our dxapp.json file are formatted as shown below:

{
  "inputSpec": [
    {
      "name": "mappings_bam",
      "label": "Mapping",
      "class": "file",
      "patterns": ["*.bam"],
      "help": "BAM format file."
    }
  ]
}

The object mappings_bam, a DNAnexus link containing the file ID of that file, will be available as an environmental variable in the applet’s execution. Use the command dx download to download the BAM file. By default, when we download a file, we will keep the filename of the object on the platform.

dx download "${mappings_bam}"

SAMtools Count

Here, we use the bash helper variable mappings_bam_name. For file inputs, the DNAnexus platform creates a bash variable [VARIABLE]_name that holds a string representing the filename of the object on the platform; because we downloaded the file using default parameters, this will be the filename of the object on this worker as well. We use another helper variable, [VARIABLE]_prefix, the filename of the object minus any suffixes specified in the input field patterns. From the input spec above, the only pattern present is '["*.bam"]', so the platform will remove the trailing “.bam” and create the helper variable [VARIABLE]_prefix for our use.

readcount=$(samtools view -c "${mappings_bam_name}")
echo "Total reads: ${readcount}" > "${mappings_bam_prefix}.txt"

Upload Result

Use dx upload command to upload data to the platform. This will upload the file into the job container, a temporary project that holds onto files associated with the job. When running the command dx upload with the flag --brief, the command will return just the file ID.

counts_txt_id=$(dx upload "${mappings_bam_prefix}.txt" --brief)

Job containers are an integral part of the execution process. To learn more see Containers for Execution.

Associate With Output

The output of an applet must be declared before the applet is even built. Looking back to the dxapp.json file, we see the following:

{
  "name": "counts_txt",
  "class": "file",
  "label": "Read count file",
  "patterns": [
    "*.txt"
  ],
  "help": "Output file with Total reads as the first line."
}

We declared a file type output named counts_txt. In the applet script, we must tell the system what file should be associated with the output counts_txt. On job completion, usually at the end of the script, this file will be copied from the temporary job container to the project that launched the job.

dx-jobutil-add-output counts_txt "${counts_txt_id}" --class=file