アブストラクト
| Date | 14:00-15:00 Nov 13, 2023 |
| Speaker | Wolfram Weckwerth Professor, University of Vienna |
| Title | Multiomics, AI and data-driven inverse modelling - from environmental sciences to molecular medicine |
| Abstract docx |
Genome sequencing and systems biology are revolutionizing life sciences.
In the last decade transcriptomics and RNAseq techniques revealed that
dynamics of mRNA represent only a small part of a complex regulatory
biochemical network which is yet unpredictable from genome sequences.
Consequently, proteomics and metabolomics emerge as cornerstone
technologies for gene function analysis and genome-scale reconstruction
of dynamic metabolic networks. Here, an integrated
proteomics/phosphoproteomics/metabolomics platform suited for functional
genomics and systems biology is presented. This platform serves also as
the basis for a recently established Vienna Metabolomics Center (VIME). A convenient workflow for data
processing, integration and mining will be presented. This strategy is
based on the data mining toolbox COVAIN (COVAriance INverse) for data
integration, multivariate statistical analysis, machine learning and
genome-scale metabolic modelling. A novel algorithm and applications
from environmental sciences up to molecular medicine for data-driven
inverse calculation of biochemical regulation from high throughput
metabolomics data implemented in COVAIN will be presented. The
algorithm has been recently extended to a fully automated workflow
integrating an automated genome-scale metabolic reconstruction (RECON)
with a novel regression loss algorithm to determine strongest causal
perturbations from metabolomics covariance networks (COV) in large
metabolic networks. Based on the combination of covariance analysis
(COV) and metabolic reconstruction (RECON) the algorithm is called
COVRECON. We applied this algorithm to the analysis of mTOR-dependent
immune system modulation. Activation of immune cells is accompanied
by a metabolic reconfiguration of their cellular energy metabolism
including shifts in glycolysis and mitochondrial respiration that
critically regulate functional effector responses. However, while
current mass spectrometry strategies identify overall or flux-dependent
metabolite profiles of cells or tissues, they fail to comprehensively
identify the checkpoint nodes and enzymes that are responsible for
different metabolic outputs. Here, we demonstrate that a data-driven
inverse modelling approach from mass spectrometry metabolomics data can
be used to identify causal biochemical nodes that influence overall
metabolic profiles and reactions. Using multiomics metabolomics,
proteomics, phosphoproteomics, transcriptomics analysis as well as
enzymatic activity measurements we identified metabolic signatures of
energy signaling and macrophage differentiation. The presented
concept of data-driven inverse modelling and multiomics analysis allows
for the systematic integration of genome-scale metabolic reconstruction,
prediction and analysis of causal biochemical regulation in microbes,
plants, animals and human.
|