home..

Prac5 6 Association Analysis In Uk Biobank

November 2022

Introduction

Practical sessions on Thursday morning and Friday morning will focus on how accelerometer-derived phenotypes can be used in UK Biobank to generate epidemiological insights.

The session on Thursday focuses on accessing the data. We will understand some of the non-accelerometer data available in UK Biobank, meet the UK Biobank Research Analysis Platform and use it to prepare data for analysis.

The session on Friday focuses on running an epidemiological analysis using data from UK Biobank.

Both sessions use this demo pipeline, with Thursday’s session focussing on notebook 1 and Friday’s on notebooks 2-4.

Tutors: Aidan Acquah, Alaina Shreves, Rosemary Walmsley

Information Governance and Security

When working with participant data, good Information Governance/Security is essential. Work with data on the Research Analysis Platform only. Do not download data onto your machine. Watch out for accidental data download e.g. a Jupyter notebook may inadvertently contain data (e.g. from printing parts of the data).

Practicalities: getting the notebooks onto the UK Biobank Research Analysis Platform

The UK Biobank Research Analysis Platform has permanent storage. In particular, you will be using a project called ‘CDT-DataChallenge-Dec22’. We’ll set up a folder under ‘users’ for each person. Please store your files in that folder. When you first log in to the platform, you may wish to set up folders called ‘data’ and ‘outputs’ within this folder (some of the notebooks will assume you have done that).

When you are running JupyterLab (or RStudio), you are temporarily running a cloud computer that has some temporary storage associated with it. You can transfer things from permanent storage to temporary storage and back again. Important: at the end of a session, before terminating your cloud computer, you need to make sure anything you need from the temporary storage is transferred over to permanent storage. If not, it will be gone!

There are different ways you could get the practical materials from GitHub onto the Research Analysis Platform. Here’s one:

I. Launch JupyterLab + Spark (as you will need for the Thursday practicals)

II. Open a Terminal instance from JupyterLab (File > New > Terminal)

III. Clone the repository by running:

$ git clone https://github.com/OxWearables/rap_wearables.git # need to use https clone as ssh clone doesn't seem to be set up

IV. Note that this has put the repository into temporary storage only. If you want to be able to access it after the session, you’ll still need to upload it to permanent storage.

V. To work with the repository, change the directory into the repository and check out the relevant branch:

$ cd rap_wearables
$ git checkout cdt_dec_2022 # This branch has some minor changes relative to main making it more suitable for our work this week 
$ cd .. # We go back up one level just so as to avoid filling the rap_wearables repo with data etc

VI. You can now run notebooks from the repository and edit them as you like.

VII. At the end of the session, you can upload to permanent storage by running (again in the terminal):

$ dx upload -r rap_wearables --dest users/Rosemary_Walmsley32/ # Remember to change my username to yours! Also don't miss the trailing slash. 
# dx is a command line client produced by DNANexus

Alternatively, if you already have things in permanent storage (e.g. having previously run through the steps above), you can download them to your temporary instance using:

$ dx download -r users/Rosemary_Walmsley32/rap_wearables # again change the file path as appropriate

Getting files between the VM and the RAP using the command line interface

This is not part of the practical. However, in the data challenge you may find you need to transfer files between the VM and the RAP. These instructions introduce how you can do that with the Command Line Interface.

Aim

These notes are for you if you:

We are going to use the command line interface to transfer the file directly and securely from the VM to the RAP.

Reasons you might want to upload a file to the RAP: You have processed the accelerometer data on BMRC (yay!) and can access a summary file containing the accelerometer phenotypes on your VM. You now want to upload it to the RAP to link it with other participant data.

Reasons you might want to download a file from the RAP: PLEASE DO NOT DO THIS IN GENERAL! But, you may want to download a text file containing participant IDs in your cohort from the RAP to your VM so you only process the subset of accelerometer data you need.

Steps

  1. Login to the VM and complete all steps below on the VM.

  2. Ensure your existing conda environment is activated by running: conda activate wearables_workshop.

  3. Run pip3 install dxpy (Downloads - DNAnexus Documentation)). You may find you need to update pip: pip install --upgrade pip.

  4. Navigate to the directory containing the files you need locally*.

  5. Run dx login to log you in to the RAP (Command Line Quickstart - DNAnexus Documentation. You will need to enter your username and password. You may also have to select the project you want.

  6. Run dx cd users/Rosemary_Walmsley32 to navigate between directories on the RAP. [This is just an example - navigate to the folder you want, not to my folder :)] You can read more about the dx command line interface in the links, but you can also run other commands prefaced by dx (e.g. dx ls)

  7. To download the file my_eid_list.txt, run dx download my_eid_list.txt.

  8. To upload the file accelerometer_summary.csv, run dx upload accelerometer_summary.csv. [Note the similarities with how you work with files in the temporary storage on a cloud instance.]

  9. Logout using dx logout. Your file should now be downloaded/ uploaded.

*This is not strictly necessary. You can navigate between directories or control it using filepaths later. But this keeps it simple.

Troubleshooting and bugs

If you have a question, speak to any of the tutors (Aidan Acquah, Alaina Shreves, Rosemary Walmsley).

There are probably bugs. If you find them, please let us know!

We’d also love to hear any suggestions for how we could improve these practicals for future students.

Lots of online resources are available to understand the UK Biobank study and the Research Analysis Platform. Some are linked to within the individual tutorials, but some more general resources are listed here. Please do spend some time exploring them - this will be useful both now and in the Data Challenge!

© 2023 Oxford Wearables Group