Resumen de A computational intelligence analysis of g protein-coupled receptor sequences for pharmacoproteomic applications
Martha Ivón Cárdenas Domínguez
Arguably, drug research has contributed more to the progress of medicine during the past decades than any other scientific factor. One of the main areas of drug research is related to the analysis of proteins. The world of pharmacology is becoming increasingly dependent on the advances in the fields of genomics and proteomics. This dependency brings about the challenge of finding robust methods to analyze the complex data they generate. Such challenge invites us to go one step further than traditional statistics and resort to approaches under the conceptual umbrella of artificial intelligence, including machine learning (ML), statistical pattern recognition and soft computing methods. Sound statistical principles are essential to trust the evidence base built through the use of such approaches. Statistical ML methods are thus at the core of the current thesis.
More than 50% of drugs currently available target only four key protein families, from which almost a 30% correspond to the G Protein-Coupled Receptors (GPCR) superfamily. This superfamily regulates the function of most cells in living organisms and is at the centre of the investigations reported in the current thesis. No much is known about the 3D structure of these proteins. Fortunately, plenty of information regarding their amino acid sequences is readily available. The automatic grouping and classification of GPCRs into families and these into subtypes based on sequence analysis may significantly contribute to ascertain the pharmaceutically relevant properties of this protein superfamily.
There is no biologically-relevant manner of representing the symbolic sequences describing proteins using real-valued vectors. This does not preclude the possibility of analyzing them using principled methods. These may come, amongst others, from the field of statisticalML. Particularly, kernel methods can be used to this purpose. Moreover, the visualization of high-dimensional protein sequence data can be a key exploratory tool for finding meaningful information that might be obscured by their intrinsic complexity.
That is why the objective of the research described in this thesis is twofold: first, the design of adequate visualization-oriented artificial intelligence-based methods for the analysis of GPCR sequential data, and second, the application of the developed methods in relevant pharmacoproteomic problems such as GPCR subtyping and protein alignment-free analysis.
