PythonSetup.md

layout	title	rank
default	Code, analysis and computing	6

Step-by-step instructions to getting started

1. Download and install Anaconda
2. Download and install git. See here for an excellent instroduction.
3. Download and install VSCode, including the AutoPilot and Remote extensions (the latter for working on the ALICE cluster)
4. Continue to setup your environment for analyzing IBL data (and select iblenv as your interpreter in VSCode)

SSH setup (locally)

• Pin the Anaconda Prompt to the Start bar, right-click on Properties, and change %HOMEPATH to the path where your code lives (e.g. C:\Users\username\Documents\code\).
• Connect your GitHub repo with SSH. Note: the university's Dell machines have a habit of looking for ssh keys in the P-Drive /p//.ssh/, whereas you probably want to keep your keys in /c//Users//username//.ssh. To change where SSH looks for keys, add an environment variable called HOME, and set it to C:\Users\username. Running ssh-keygen as a test, it should then prompt you to save your keys on the C-drive. If you've already created your ssh keys and they are on the P-drive, simply move them.
• MobaXTerm is your best bet for using ssh on Windows (for transferring data).

ALICE setup

• See here to connect using ssh, and here for Windows-specific instructions with custom keynames.
• Warning: your home directory only has 15GB of space. This fills up quickly if you install your conda environments there (just iblenv is 2.5GB!). So before you create your first conda env, run mkdir ~/data1/.conda; ln -fs data1/.conda. Keep only code in your home directory, and everything else (data, figures) in the lab's shared project space or your own data folder.

Python and data analysis references

• Know your way around the command line
  • Basic cheatsheet
• Understand what a virtual environment is
• Understand Git and GitHub
  • GitLectures
  • Git primer by Brad Voytek
  • Even better Git intro
  • Further practice: pull this wiki repo, add something in the Markdown language, and submit a pull request (see Home for instructions on how to contribute to the wiki).
• Learn the basics of Python
  • DataCamp has very good adaptive programming courses. Sign up with your @umail.leidenuniv.nl address using the link for free access.
  • Lab in cognition & perception, both by Todd Gureckis.
  • Students at Leiden University can follow the Introduction to Python course, offered by LIACS.
  • You can also access DataCamp learning resources with your Leiden Uni account.
• Think about the structure of data
  • The structure of data, by Todd Gureckis
  • Learn to work with the pandas and seaborn packages for data handling and visualization.
  • Pandas tutor with helpful dataviz.
• Once your code runs, make it better!
  • The Good Research Code Handbook is a fantastic guide to write elegant, organized code that doesn't make you want to tear your hair out. Highly recommended.
  • Dan Larremore's lab guides to clean code
  • Consider code review

Open data, open code, open science

• Might there be a dataset or a tool out there that does what you need? Check out this list of lists.

Computing resources

For most projects (especially those using behavioral data), your laptop will be more than sufficient to run Python . If you need more heavy lifting, there are a few options:

• ALICE supercomputer @ Leiden Uni
  • Get an account, and include a request to be added to the data_pi-uraiae project space (for IBL data).
• LISA / Cartesius clusters @ SurfSara
  • Apply through NWO. Very well managed, but since Leiden does not have a contract with SurfSara you have to apply to extend your account every year.