dlim 1.1.0

Direct Latent Interpretable Model (D-LIM): An interpretable neural network for mapping genotype to fitness.

D-LIM (Direct-Latent Interpretable Model)

Overview

D-LIM (Direct-Latent Interpretable Model) is a neural network that enhances genotype-fitness mapping by combining interpretability with predictive accuracy. It assumes independent phenotypic influences of genes on fitness, leading to advanced accuracy and insights into phenotype analysis and epistasis. The model includes an extrapolation method for better understanding of genetic interactions and integrates multiple data sources to improve performance in low-data biological research.

System Requirements

Hardware requirements

D-LIM requires only a standard computer with enough RAM to support the in-memory operations.

Software requirements

This package is supported for Linux. The package has been tested on the following systems:

• Linux: Ubuntu 20.04

Python Dependencies

D-LIM depends primarily on pytorch, as well as the components of the Python scientific stack:

• pandas
• numpy

Installation guide

• Install the package from Pypi:

pip install dlim

Or install it from the sources:

git clone https://github.com/LBiophyEvo/D-LIM-model.git
cd D-LIM-model 
pip install -e .

Documentation

The official documentation with usage is available at: https://d-lim.readthedocs.io/en/latest/. Documentation covers instructions for running D-LIM on simulated and experimental data, including demonstrations on some simple datasets.

Data avaibility

• Simulated data: see src_simulate_data\
• Experimental data:
  • Data from Kemble et al. can be found in data.
  • Data of protein-protein interaction is from Diss et al..
  • Data of yeast with different mutations and enviroments is from Kinsler et al.. You can download data from our Google driver link: Data used in D-LIM.

Manuscript reproduction

Source code to reproduce the analysis of the D-LIM manuscript are available at reproducibility folder in D-LIM GitHub.

License

This project is covered under the MIT License