ML in Omics

Project Topic: Machine Learning in Omics: Integration of Metagenomics and Metabolomics.

Source code and documentation for Life Science Technologies Project Course B, Aalto University, Finland.

How to Run

Environment Setup

(Optional) First, create a virtual environment:

python -m venv mlomics         # or `virtualenv mlomics`
source mlomics/bin/activate    # to activate the env

Next, install the requirements:

pip install -r requirements.txt

Analysis setup

First, the configuration for the study must be available in the configuration file. The config for each study looks something like this:

'mars_ibs_lr': {
    'dir': 'MARS_IBS_2020',
    'preprocessor': PC,
    'preproc_kwargs': {'var': 0.9},
    'classifier': LogisticRegression,
    'classifier_param_grid': {
        'penalty': ['l1', 'l2'],
        'C': [0.001, 0.01, 0.1, 1, 10, 100],
        'solver': ['liblinear', 'saga'],
        'max_iter': [100, 200, 300],
        'tol': [0.0001, 0.001, 0.01],
        'class_weight': [None, 'balanced'],
        'random_state': [42]
    },
},

Then, you can run the analysis using the following command and arguments:

python main.py --path "path/to/data" --study "mars_ibs_lr" --output "../output" --debug True

Analysis Description

Goals

• Integration of Metabolomics and Metagenomics data for more accurate biomarker identification
• Comparison of Metabolomics and Metagenomics in disease status prediction

Summary

Aspect	Value
Model	Logisticregression, RandomForest, ML-based classifiers (future)
Feature Engineering	PCA, LDA, $\chi^2$
Biomarker detection	Graph Convolutional Networks, model-based (LR)

Data

The datasets were obtained from this repository.

Authors

Alexandra Gorbonos

Alireza Tajmirriahi

Mohammad Sherafati

Salma Rachidi