Forum Numerica - Analysis of large scale biomedical data: Genomics and Imaging- Genetics in brain disorders


  • Marco LORENZI (Inria Epione team)
  • Boris GUTMAN (Medimaging research center, Illinois Institute of Technology, USA)
  • Barbara BARDONI (IPMC, CNRS)
  • André ALTMAN (Centre for Medical Image Computing, University College London, UK)




14h00 : Combining images and genes to better understand brain disorders

by Andre Altmann, Centre for Medical Image Computing, University College London, UK 

In this talk I will speak about imaging genetics and its application to brain disorders, in particular Alzheimer’s disease. In imaging genetics measures derived from (neuro-)imaging are used as endpoints in statistical analyses of genetic data. This is in contrast to conducting such studies with dichotomous case-control labels. Replacing crisp diagnostic labels with imaging marker that are thought to better reflect the disease process are expected to lend increased statistical power to such analyses and also allow to investigate different aspects of the underlying disease process.

In this talk, we will briefly revisit conceptual challenges that are posed by jointly analyzing genomic and neuroimaging data. I will introduce the use of imaging genetics in cases where there are no matched data, i.e., imaging and genetic data from the same subjects. Rather, synergies between data sources are generated by working with data in a common neurological reference space. This concept will be showcased by combining brain atrophy maps with gene expression data from the Allen Brain Atlas in Frontotemporal Dementia, Alzheimer's Disease and in elucidating the genetic basis of functional brain networks.

14h30 : Characterization of a cohort of patients affected by a syndromic form of autism characterized by early onset schizophrenia

by Barbara Bardoni, IPMC, CNRS, Sophia Antipolis, France 

Early Onset Schizophrenia (EOS) is a psychiatric disorder characterized by a wide range of symptoms such as delusions, hallucinations and abnormal social behavior. This disease is frequently accompanied by Autism Spectrum Disorders (ASDs) and Intellectual Disability (ID), which are believed to share a common genetic background with EOS. During this study, the Whole Exome Sequencing (WES) of 9 TRIO (child and the two parents belonging to a cohort of 40 families sharing a similar phenotype) is performed in order to find novel genes involved in the pathogenesis of EOS (McCarthy et al., 2014; Hommer and Swedo, 2015). Probands are diagnosed with EOS, which in some cases was combined with other phenotypes such as ASDs or ID. Additionally, two single cases of patients diagnosed with schizophrenia were analyzed in the course of this study. We found some very rare inherited coding variants that have a putative pathological impact on protein function according to in silico analysis. To further evaluate the pathological impact of the identified variants, we are using different approaches that depend on the characteristics and potential functions of the candidate genes. For instance, for one gene we created a cellular model in the SH-SY5Y cell line by the CRISPR-Cas9 technique mimicking the mutation found in the patient. To characterize the phenotype of the mutated cell line, we are performing the analysis of gene and protein expression and MEA analysis (Multielectrode Array). This technique - that will be carried out in collaboration with Dr. Sergio Martinoia (University of Genova, Italy) - allows the simultaneous recording of optical and electrical signals of cell activity, thus enabling the studies of the different activity patterns in the cell network.  For the other genes that have their fly homolog, we are generating a model in D. melanogaster whose power as a model in the study of human genetic diseases has been widely validated (van der Voet et al., 2014; Ugur et al., 2016).

15h30 : Statistical learning for the analysis of imaging, genetics and biological data

by Marco Lorenzi, Epione Research Project, Inria Sophia Antipolis, France

In this talk I will give an overview of recent multivariate statistical methods for the joint analysis of heterogeneous information in Alzheimer’s disease, with application to imaging-genetics and disease progression modelling. The increasing availability of large collections of medical data offers a tremendous opportunity for the development of statistical approaches for the joint modelling of the relationship between brain imaging, genetics, and clinical biomarkers, to improve the understanding of neurodegenerative disorders, as well as for better prediction and quantification of the pathology in unseen individuals. 

I will introduce multivariate statistical approaches for the analysis of the relationship between large arrays of genetic variants and brain atrophy measured by magnetic resonance imaging. This framework allows the identification of meaningful genetic locations associated to cortical and sub-cortical atrophy in Azheimer’s disease, and provides novel insights for the testing of hypothesis about the biological mechanisms of the pathology. I will also illustrate novel probabilistic methods for the estimation of the evolution of biomarkers from time-series of individual’s biomarker measurements in clinical trials. Thanks to these approaches we can provide a biologically plausible statistical description of the evolution of Alzheimer’s pathology across the whole disease time-span, as well as remarkable diagnostic performances when tested on a large clinical cohort in a clinical trial setting. 

16h00 : Understanding Genetic Control of the Brain with MR Imaging

by Boris Gutman, Med imaging research center, Armour College of Engineering, Illinois Institute of Technology, USA

 The ability to perform sequencing of the human genome has opened up the possibility to quantitatively assess genetic control of virtually any measurable trait. In practice, this is hampered by insufficient sample sizes and for certain problems, the high dimensionality of the phenotype. Indeed, this has until recently been the case with traits extracted from brain MRI. However, recent efforts to collect large harmonized brain MRI datasets have made genome-wide studies of brain imaging traits possible. In this talk, I will show some of the computational techniques and results used for extracting reliable brain MR measures, validating their heritability and finally discovering specific genetic variants associated with the brain.
About the speakers

Marco Lorenzi - is a junior research scientist at the French Institute for Research in Computer Science and Automation (Inria) and permanent research member of the Epione Research Group of Inria Sophia Antipolis. After his studies in Mathematics at the University of Turin, Italy, in 2007 he was research assistant at the Laboratory of Neuroimaging of the clinical institute Fatebenefratelli, Brescia, Italy. He obtained his PhD in 2012 at Inria Sophia Antipolis, and was subsequently Research Associate in the Centre of Medical Image Computing (CMIC) of University College London (UCL), UK. His research activity concerns the development and study of computational and statistical methods for the analysis of biomedical data and medical images. 

Barbara Bardoni - after her studies in Genetics at the Faculty of Sciences of the University of Pavia (Italy), Dr. Barbara Bardoni got a position of Assistant Professor at the Faculty of Medicine of the same University. She integrated the team of Prof. J-L. Mandel at the IGBMC in Strasbourg in 1997. In 2002 she was recruited at INSERM as Researcher and in 2007 she was nominated Research Director. She currently heads the team “RNA metabolism and neurodevelopmental disorders” at the CNRS Institute of Molecular and Cellular Pharmacology (Sophia-Antipolis - Valbonne). Since 1997 her main interest is represented by the molecular and cellular bases of the Fragile X Syndrome, the most common form of inherited intellectual disability and autism, and the search for a therapeutic approach for this disorder. Her researches have been supported at the international level (European Community, NIH-USA, Human Frontiers Science Program and NIMH-Australia).

Andre Altmann - studied Computer Science at the RWTH Aachen and graduated in 2005 with a work in the field of spoken language recognition at the Chair for Computer Science 6. He pursued his PhD studies in Saarbrücken at the Max Planck Institute for Informatics in the Computational Biology group of Thomas Lengauer. From February 2010 till May 2012 he was a Postdoctoral researcher in the Statistical Genetics Group headed by Bertram Müller-Myhsok at the Max Planck Institute of Psychiatry in Munich. From July 2012 till July 2015 he worked at the FIND lab of the Stanford University headed by Michael D Greicius first as a Postdoctoral Scholar (till January 2015) and later as an Instructor. In August 2015 he joined UCL‘s Centre of Medical Image Computing (CMIC) as a MRC Senior Fellow with support from the eMedLab project.

Boris Gutman - Boris Gutman is professor of biomedical engineering at the Illinois Institute of Technology. He received his PhD in biomedical engineering at the University of California Los Angeles and postdoctoral training in imaging genetics at University of Southern California. In his research, he develops computational tools modeling brain morphometry and connectivity, and applies these tools to the study of brain disease and its genetic underpinnings.