omics-integration

Multi-omics data integration uses unsupervised factor analysis like MOFA+ to identify latent factors explaining shared variation across genomics, transcriptomics, and proteomics layers. This reveals cross-modal regulatory relationships and patient subgroups from paired cohort studies.

Supervised integration via DIABLO builds predictive models to identify biomarkers correlating across multiple data blocks. This approach maps cross-modal regulatory cascades by linking RNA-seq and proteomics data in paired clinical cohorts.

To perform MOFA+ analysis, input your RNA, protein, and metabolite datasets into the skill to compute latent biological factors. The process uses the mofapy2 library to extract shared variance across the different omics layers.

Yes, advanced supervised modeling and classification require R-based mixOmics integration. While standard scientific Python libraries like pandas, numpy, and scikit-learn handle unsupervised analysis, DIABLO models specifically depend on the mixOmics framework.

Yes, network-based correlation modeling constructs cross-omics correlation networks and performs patient stratification using Similarity Network Fusion (SNF). This leverages networkx to model complex biological relationships across heterogeneous molecular datasets.

Multi-omics factor analysis and similarity network fusion require standard scientific Python libraries including mofapy2, pandas, numpy, scipy, networkx, scikit-learn, statsmodels, matplotlib, and seaborn for data handling and visualization.