- This event has passed.
Insigneo Seminar: Data Science meets computational fluid dynamics – developing new tools to interrogate the mechanosensitive epigenome of endothelial cells during plaque development in vivo
Friday, 15 March, 11:30 am - 12:30 pm GMT
We are delighted to announce that Professor Rob Krams, Professor in Molecular Bioengineering at Queen Mary University of London will be visiting the Insigneo Institute on Friday 15th March and will give a seminar on ‘Data Science meets computational fluid dynamics – developing new tools to interrogate the mechanosensitive epigenome of endothelial cells during plaque development in vivo‘ at 11.30am – 12.30pm in Lecture Theatre 3, The Medical School, Royal Hallamshire Hospital.
Data science meets Computational Fluid Dynamics: Developing new tools to interrogate the mechanosensitive epigenome of endothelial cells during plaque development in vivo
Yean Kok Chooi 2, Roza Nikolopoulou2, C.A. Mein3, Eva Wozniak3, Alex Delahunty2, Nasar Jarka1,2,Miten B. Patel1,2, Fotios Savvopoulos1,2, Pan Jang1,2, Ranil de Silva1, and Rob Krams2.
1NHLI, Imperial College, and 2 Bioengineering and 3Genomics Centre, Queen Mary’s and the Genome Centre, Queen Mary University
We have developed novel methods to induce (stent/cuff), monitor (micro-CT, IVUS, OCT, angio), 3D reconstruct and calculate (CFD, FSI modelling) pro-atherogenic biomechanical fields (disturbed endothelial shear stress and strain) in mice carotid, pig and human coronary arteries and were able to identify disturbed biomechanical patterns in all species predicting advanced atherosclerosis using state-of-the art machine learning (SVM and Neural Networks) modelling techniques.
In order to develop novel interventions, we explored the underlying molecular signatures using laser capture techniques to isolate total RNA of endothelial cells exposed to these pro-atherogenic
biomechanical patterns. Deep sequencing was, subsequently used to identify differentially expressed genes and their epigenetic control.
After extensive Q&A, 2300 differentially expressed mechanosensitive mRNA’s were identified encompassing >50 signaling pathways consisting of known (MAPK, NF-kB) and unknown signaling
(Insulin, WnT, TGF, and Notch) pathways. In a parallel sample, 1,500 microRNA (miR’s) were identified of which ~50% were differentially expressed. These differentially expressed miR’s were
either sub-divided into structurally defined families using miRBASE or bicistronically co-expressed miR’s using sparse-PCA. MiR-families – which regulate similar genes – were only partially co-
expressed (~30%), leaving the remaining co-expressed miR’s (~70%) as potential regulators of signaling pathways.
These hypothesis was tested using either Elastic Net to find dominant miR-mRNA interactions or using partial least squared to find more subtle miR-mRNA interactions. Indeed, miR-families, were either regulating similar genes through co-expression or through differentially expression. The non-family co-expressed miR’s which are regulating multiple genes, are currently studied for their role in signaling pathways.
The validity of the bioinformatics approach, and the cooperation between family and non-family members is tested with a novel high throughput CRISPr-platform where >800 independent
interventions will be performed to evaluate above formulated hypothesis.
In conclusion, we have developed and are developing novel methods to identify predictive biomechanical models for the development of atherosclerosis. In order to provide new interventions, we are studying the molecular signature of endothelial cells exposed to these patterns and as mechanotransduction is too complex we focus on miRNA as a novel way of intervention. All methods use an interplay between finite element modeling and data science based machine leaning methods.
Thanks to the British Heart Foundation for financial support (RG/11/13/29055 and PG/15/49/31595).