Pythia Biosciences to advance multi-omics analysis with state-of-the-art ML and workflows

“Integrated multi-omics analysis” has become an exciting pillar of bioinformatics in the last few years. Stacking data from multiple “omes” (genomes, epigenomes, transcriptomes, proteomes, metabolomes), this approach enables scientists to have a more complete picture of the biological systems and disease mechanisms. However, multi-omics analysis is computationally challenging by its nature, which involves many problems such as batch effects, data and metadata inconsistency, as well as keeping up with new methods developed every day!

To help scientists overcome these challenges and harness the power of multi-omics, Pythia Biosciences developed a new SAAS platform called CDIAM Multi-Omics Studio (https://www.c-diam.com ), incorporating state-of-the-art machine learning and workflows in an easy-to-use graphical UI. Standing on a microservice-based architecture, it allows easy integrations of new analysis methods.

“Pythia’s mission is to help patients by helping scientists interpret omics data so for the majority of our first year we focused on the fundamentals of omics data visualization and analysis, pathway and network analysis, target, biomarker and mechanistic workflows. Pythia’s AI strategy is to bring well tested and benchmarked machine learning algorithms to scientists while at the same time building powerful feature sets and taking advantage of the advanced AI tools available from our cloud partners like SAGE from AWS for example.” said Tristan Gill, the CEO of Pythia Biosciences.

What’s supported:

• Predict biomarkers and DE genes with “Extreme Gradient Boosting” (XGBoost), a variant of the Gradient Boosting Machine (GBM), a machine learning classifier developed by Chen et al. that has been widely applied for classification problems and predict biomarkers in biology.
• Build the pipelines for finding differentially expressed genes, proteins, and differentially accumulated metabolites; pathway enrichment analysis, target and biomarker prioritization; cell-cell communication; gene co-localization; network algorithms for mechanism of disease or drug target actions and many more.
• Interactively visualize single-cell omics data.
• Summarize important aggregate insights, such as common significant pathways; scoring and identifying most potential targets and biomarkers; finding most significant cell-cell interactions and active ligand-receptor pairs – across different omics experiments.
• Access a variety of public databases to validate results across omics, such as spatial databases (10X Visium, Nanostring CosMx SMI and GeoMx DSP,…), Recount3 RNA-seq data and NCI Proteomics Data Commons.

Input data can be any of the following: mutation data, ATAC-seq, bulk RNA-seq, scRNA-seq, mass-spectrometry proteomics, metabolomics, spatial omics,…

The platform is currently available for a trial upon request at https://c-diam.com .

Learn more about CDIAM and Pythia Biosciences at: CDIAM: A multi-omics analysis software platform with adaptability and ease of use (pythiabio.com)