Running lobSTR v.3 using Arvados¶

This tutorial demonstrates how to run the lobSTR pipeline using the example that the Erlich Lab provides at their page. LobSTR is a tool for profiling Short Tandem Repeats (STRs) from high throughput sequencing data. The lobSTR publication is available here: Gymrek M, Golan D, Rosset S, & Erlich Y. lobSTR: A short tandem repeat profiler for personal genomes. Genome Research. 2012 April 22. This tutorial introduces the following Arvados features:

• How to run lobSTR v.3 using Arvados
• How to access your pipeline results.
• How to browse and select your input data for lobSTR and submit re-run the pipeline.

1. Start at the Curoverse website and click Log In at the top. We currently support all Google / Google Apps accounts for authentication. By simply choosing a Google-based account, your account will be automatically created and redirect to the Arvados Workbench.
2. In the Active pipelines panel, click on the Run a pipeline... button. Doing so opens a dialog box titled Choose a pipeline to run.
3. Select lobstr v.3 and click the Next: choose inputs button. Doing so loads a new page to supply the inputs for the pipeline.
4. The default inputs from the lobSTR source code repository are already pre-loaded. Click on the Run button. The page updates to show you that the pipeline has been submitted to run on the Arvados cluster.
5. After the pipeline starts running, you can track its progress by watching log messages from jobs. This page refreshes automatically. You will see a complete label under the job the column when the pipeline completes successfully. The current run time of the job in CPU and clock hours is also displayed. You can view individual job details by clicking on the job name.
6. Once the job is finished, the output can be viewed to the right of the run time.
7. Click on the download button to the right of the file to download your results, or the magnifying glass to quickly view your results.

Uploading data and using it on Arvados¶

Full documentation can be found here

1. Install the Arvados Python SDK on the system from which you will upload the data (such as your workstation, or a server containing data from your sequencer). Doing so will install the Arvados file upload tool, arv-put.
2. To configure the environment with the Arvados instance host name and authentication token, see here
3. Navigate back to your Workbench dashboard and create a new project by clicking on the Projects dropdown menu and clicking Home.
4. Click on [+ Add a subproject]. Feel free to edit the Project name or description by clicking the pencil to the right of the text.
5. To add data, return back to your shell and create a folder and put the two paired end fastq files you want to upload inside. Use the command arv-put * --project-uuid qr1hi-xxxxx-yyyyyyyyyyyyyyy. The qr1hi tag can be found in the url of your new project. This ensures that all the files you would like to upload are in one collection.
6. The output value xxxxxxxxxxxxxxxxxxxx+yyyy is the Arvados collection locator that uniquely describes this file.
7. Once that is uploaded, navigate back to the dashboard and click on Run a pipeline... and choose lobstr v.3.
8. You can change the input by clicking on [Choose] next to the Input fastq collection ID.
9. Click on the Dropdown menu and click on your created project and choose your desired input collection. Click OK and Run to run lobSTR v.3 on your data!

FAQ¶

• Does this support both paired-end and single-end reads?
  • Currently, the pipeline template only supports paired end reads. If you would like to run a single-end read experiment, please email support@curoverse.com and we will make a custom pipeline template to do so. You can also feel free to copy the template yourself and edit the commands! Documentation is provided here

• What type of files does this support?
  • It supports and fastq files with a variety of names as long as they have the string "1.f" and "2.f". (.fq, .fas, and .fastq are all supported).

• Can this run multiple samples at once?
  • We are currently working on supporting parallelization of running batch processing on multiple samples, and it should be ready soon.