We are the group of Julio Saez-Rodriguez, a research group at the Heidelberg University Hospital. We develop software tools for systems level analysis and mechanistic modeling of molecular and biomedical data.
Our goal is to acquire a functional understanding of the deregulation of signalling networks in disease and to apply this knowledge to develop novel therapeutics. We focus on cancer, auto-immune and fibrotic disease. Towards this goal, we integrate big (‘omics’) data with mechanistic molecular knowledge into statistical and machine learning methods. To this end, we have developed a range of tools in different areas of biomedical research, mainly using the programming languages R and Python.
Database knowledge is essential for many of our methods, to this end we
maintain our own database called OmniPath, combining data from about
150 original resources. OmniPath consists of a database builder Python
module (pypath), a public web service, and web service clients for R,
Python and Cytoscape. Read more about OmniPath in our paper, and
check out the repos below.
OmnipathR: R client for the OmniPath web service- Python client for the OmniPath web service
- Cytoscape client
pypath: database builder Python module for OmniPathBioCypher: graph database (Neo4j) framework for molecular databasesDoRothEA: a comprehensive gene regulatory network
We have a comprehensive tutorial presenting a typical functional omics workflow. It uses footprint methods to infer transcription factor and pathway activities from transcriptomics data, and uses causal reasoning to infer the upstream pathways based on these activities.
Our functional omics pipeline is also available as a web application called FUNKI. Check out the app here, find the code in the git repo, and read more in our preprint.
PROGENy stands for Pathway RespOnsive GENe activity inference. It is a footprint method to infer activities of canonical signaling pathways based on gene expression data. Read more in the repos below and in our paper.
DoRothEA is a comprehensive gene regulatory network compiled from literature, TF binding site prediction, ChIP-Seq and gene expression based inference methods. It is also a footprint method to infer transcription factor activities based on gene expression data, using the VIPER algorithm. Read more in our paper and in the repos below. DoRothEA data is also available in OmniPath.
decoupleR is a unified framework to extract functional signatures from omics data. Read more in our preprint.
- R package
- Python module (not yet public)
CARNIVAL uses causal reasoning to find the most plausible networks that explain the activity patterns derived from omics data. See details in our paper.
- R package
- CARNIPHAL: R package for running CARNIVAL with phosphoproteomics data
- Python module: coming soon!
COSMOS applies causal reasoning (using CARNIVAL) for the combined analysis of multiple omics data: phosphoproteomics, transcriptomics and metabolomics. See more in our paper, the video tutorial and the repos below.
CellNOpt creates predictive, mechanistic models of signaling networks using logic formalisms (Boolean, fuzzy, or differential equations) based on the combination of prior knowledge and omics data. Read our paper and check out the CellNOpt landing page and the repos below.
PHONEMeS trains Boolean models on phosphoproteomics data. See more in the PHONEMeS page, the papers (ILP implementation, original) and the repo below.
LIANA is a framework combining a number of ligand-receptor database resources with a number of methods to infer ligand-receptor activities from single-cell transcriptomics data. More details in our manuscript and the repos below.
MISTy leverages on explainable machine learning to find relationships between the expression of various markers across a number of spatial contexts (views; e.g. intrinsic, juxtacrine, paracrine). Read more in our preprint and the repos below.
The enzyme enrichment analysis is a footprint method for inferring enzyme activities from metabolomics data. It is implemented in the OCEAN package.