problems with "ancestry reports"
11:45am-12:45pm PST Friday August 8, 2019
Have you ever wondered how "race" and ancestry are "calculated" by 23andMe or Ancestry.com? Have you ever questioned why your results seemed so "wrong"? Well, in this workshop we'll be building our own admixture (ancestry) reports from publicly available data as well as showing how we can "hack race" to make racial results different.
The respository is based on modified code from https://github.com/hagax8/ancestry_viz
Make sure you have git, python, and pip installed
Step 0 (Optional): Make a virtual environment
This part is optional but encouraged because we're about to download files and stuff that might mess up existing dependencies you have in your dev environment. We'll be following the instructions from: https://docs.python-guide.org/dev/virtualenvs/
$ cd ~
$ sudo apt install virtualenv
$ cd <directory_path_you_work_in> ie.
$ cd Downloads or something
$ virtualenv -p /usr/bin/python2.7 venv to make a virtual environment named
$ source venv/bin/activate to activate the virtual environment
Step 1: CLONE THIS REPO
$ cd venv
$ git clone https://github.com/herroannekim/hacking_race.git into your virtual environment
Step 2: Download the Data
Step 3: Install dependencies
Either make a virtual environment or install the dependencies to your system:
$ pip install -r requirements.txt
Step 4: Run ancestry prediction and clustering
In this example, we're running a prediction on sample HG01108, a Puerto Rican female
$ python2 hacker.py --model GTM --out classify_HG01108 --config ./standard_config.json --classify-id HG01108
Step 5: Hack 3:D
For the ancestry hack, we will now shift her ancestry towards Kenyan using
$ python2 hacker.py --model GTM --out kenya_dig_it --classify-id HG01108 --config ./standard_config.json --manipulate-towards "Luhya_in_Webuye,_Kenya"
Step 6: Check your work
- Now you should be able to check the output and cluster graphs :)
Step 7 (optional): Clean up
If you used a virtual environment, you can clean up with