Signal processing and machine learning for omics data

Group: Signal and information processing for sensing systems
Group leader: Santi Marco (

In the realm of omics data, current analytical instrumentation applied to biomedical fluids or food samples offers the possibility to analyse thousands of analytes in parallel. Current instrumentation such as Nuclear Magnetic Resonance and Mass Spectrometry in untargeted scenarios produce very rich and complex data whose interpretation is far from trivial.

This project aims to the design, implementation and optimization of full workflows for the analysis of these data that go from raw signals to metabolite abundances.  These workflows consist of several steps that require the application of state-of-the-art techniques in signal processing, machine learning, statistical methods, optimization techniques, etc.

We search for a PhD candidate with passion for signal processing and machine learning techniques applied to challenging biomedical data.  Research in this area will introduce the student to the latest methods in signal pre-processing with heavy use of wavelet methods and tensorial signal processing for blind source separation. In the machine learning part, we will explore the latest trends in feature selection methods for biomarker discovery using sophisticated combinations filters and wrappers based in linear and non-linear methods such as random forests and support vector machines. Emphasis will be made on the integration of multiple omics datasets using techniques such as canonical correlation regression and orthogonal partial least squares.

Data Analytics will be done mostly in R technology. The candidate should have a technical background in engineering or bioinformatics with previous experience in programming and a passion for algorithmic integration and test.