Project: Predicting Phenotype through Interaction of Genotype, Epigenotype and Environment with Probabilistic Models
Max Planck Institute of Psychiatry
Ilaria Bonavita was born in Brindisi, Italy, in 1991. She studied at the Polytechnic University of Turin and obtained her Msc in 2015 working on Automatic Requirement Classification Tool Based on Machine Learning Techniques and Statistical Approaches.
Deep Learning is a growing new area of Machine Learning which has been introduced with the aim of discovering intricate structures and inner relations in large data sets. A number of deep learning architectures have been developed so far and applied to various machine learning problems ranging from feature selection and classification to pattern recognition and prediction. These recently developed techniques have high potential in facing many of the key challenges of statistical genetics.
The first phase of my project will consist of a scouting of existing implementations and methods in deep learning. Algorithms and techniques primarily found to be of interest will then tested on a first gene-expression dataset. A larger and updated version of the dataset will be available soon, allowing for more complex experiments. A second dataset will also become available during 2016.
- Primarily, the objective will be to prototype a model that allows the exploration of the interaction between environment and genotype as it effects phenotype.
- Then, the model would be augmented with epigenomic data in order to investigate the ability to interact the epigenomics with genotype.
- In a third step we will augment the model with transcriptomic data.
Computations likely will be heavy and we will use the comparatively large infrastructure of graphics cards computing systems at the Max Planck Institute of Psychiatry to ease the computational burden.
Eventually, this would lead to the creation of a full model to be available to the scientific community.
Motivation for participating in the network
I have always been persuaded of the idea that a multidisciplinary approach to every kind of problem provides useful and deeper insight into the problem itself. My academic training has been indeed extremely cross-disciplinary as it places at the border between Mathematics and Engineering. In the last year of my Master I became particularly concerned with the issues related with Big Data and with the possibility of investigating them combining traditional Statistics with newly developed Machine Learning algorithms.
The MLPM Marie Curie program seemed to me a great opportunity to gain wider understanding of Machine Learning and to finally apply the Mathematical formalism and methodology I leaned to a scientific topic of recent interest like Personalized Medicine.
One of my interests is also to strengthen my understanding of biology and genetics and thanks to this program I will benefit from the cooperation with experts and institutes in these fields.
Lastly, thanks to summer schools, seminars and research activities involving all the partner nodes, the program promotes a multicultural, cooperative and creative environment to work in.
Duration of fellowship: from November 2015 to December 2016
- MLPM2012 has ended – thank you!
- Final ITN meeting and Krupp Symposium in Munich
- MLPM ITN fellow Melanie brings science to classrooms and inspires with simple but exciting experiments
- ESHG Symposium – a great success!
- Team working event: The 2nd ITN March retreat
- January (1)
- Awards (2)