About

The Pipeline of Interpretable Classification for Omics (PICO) is a tool to democratize machine learning (ML) for omics data.

It must be installed on a computer and possesses a visual interface which can be used to perform an ML experiment, from uploading the data to analysing the results. Advanced users will find that a script automate.py allows the steps from uploading data to training the algorithms to be run directly with a terminal (command prompt). The output file can then be loaded into the PICO interface for results analysis. Expert users will find that the automate script can be run on a distant server (need to : install the environment on the server, understand nodes/parallelization, know how much resources it will need to run.)

Throughout the interface, when clicking on something, if the expected effect is not displayed immediately, check the title of the web navigator tab. If an action has indeed been triggered it will be written “Updating…”.

License
Metabolomic Dashboard for Interpretable Classification (MeDIC) © 2025 by Elina Francovic-Fontaine is licensed under CC BY-NC-SA 4.0

Requirements

The recommended configuration considers the user will need the computer while the experiment is runing.

Minimal requirements Recommended requirements
8 threads/core processor 12 threads/core processor
8GB RAM 16GB RAM
Parallelization

Enabling multithreading allows PICO to use all the performance of the CPU. It reduces dramatically the compute time of the experiment. The speed increase depends on the CPU’s performances (number of threads/core and speed).

Benchmarks

Here are some example of machine setups vs compute times. They are all refering to the experiment described in the paper of all controls vs all cases and are all using parallelization. Many things can factors into the variations of speed, like the cache memory, the speed of the processor (Hz), the use of the computer while the experiment is running and more. The following benchmarks are simply to give an overview with simple references.

Linux
  1. Debian 12 with 16gb RAM and 8 threads : ~35 minutes
  2. In WSL, Ubuntu with 8 gb RAM and 12 threads : ~40 minutes
  3. In docker container, Fedora with 6gb VRAM and 12 threads : ~15 minutes
Mac
  1. MacOS 14.6.1 with 16gb RAM and 8 core : ~10 minutes
  2. MacOS 15.5 with M4 chip 32gb RAM and 10 core : ~5 minutes
Windows
  1. Windows 10 version 10.0.19045 with 16 gb RAM and 8 core : ~15min