Title: Inference and revision in Biological regulatory networks

Date: 02/04/2018

Time: 16h30

Location: Room 0.20, Pavilhão de Informática II, Alameda

Advisors: Prof. Pedro Tiago Gonçalves Monteiro / Prof.ª Maria Inês Camarate de Campos Lynce de Faria 

Abstract: Cellular processes are governed by complex regulatory networks consisting of genes, proteins, and their interactions. Modelling such processes is a crucial step towards the formal understanding of the interaction between cell components. These models can be quantitative or qualitative models. However, for many biological processes, there is no detailed quantitative information. Qualitative models have proven to be well adapted for the modelling of systems where quantitative information is generally incomplete or noisy. Qualitative models consider a small set of possible values for each component, e.g., active/inactive. There are several works over biological networks using different qualitative formalisms. Some of the areas of work are reasoning over biological networks, checking the consistency between a model and a set of experimental data, and finding attractors of a network and their reachability. Nevertheless, constructing biological models is still mainly a manual task performed by a modeller, typically an expert in the domain. As new data is acquired, a model needs to be revised or updated. The new data may not be consistent with the current model. Approaches to model revision relying on the Sign Consistency Model (SCM) have been implemented using logic-based tools. However, the SCM lacks in expressiveness for regulatory functions. The logical formalism, which has more expressiveness regarding regulatory functions, has been successfully used and implemented using logic basedtools, such as Answer Set Programming, and Boolean satisfiability. The goal of the proposed work is to extend the current approaches of model revision using SCM to the logical formalism. The first phase will focus on consistency-checking, implementing the technique over different logic-based tools in order to compare different approaches. Then, in case of inconsistency, the goal is to provide explanations for the inconsistency and, if necessary, rank them in a biological meaningful way, to be presented to a modeller. Finally, considering the most plausible cause for inconsistency, a revision process must be defined, as a combination of atomic operations. In the end, the work should result in a procedure and correspondent tool capable of semi-automatically building a consistent model iteratively as new data is acquired.