R Costa

LAQV, DQ, Faculty of Science and Technology, NOVA University Lisbon

e-mail: rs [dot] costa [at] fct [dot] unl [dot] pt

Short Bio: R. Costa is currently an assistant researcher at FCT-NOVA. He is also an invited assistant professor. His research focuses on systems biology modeling and data science with application in biotechnology, (bio)processes and healthcare.

Research Interests

The main topic of my PhD was in metabolic modeling, control and optimization of biological systems — a focus that continues to drive my research. Since my PhD, I have been interested on the development and application of computational approaches, algorithms, models, and databases to Systems Biology and Biotechnology, particularly for system level analysis of microbial metabolism. To this end, I employ different modeling approaches including ODE-based, constraint-based, data-driven and hybrid approaches in close collaboration with experimental groups. I also belong to a consortium of researchers interested in whole-cell modelling using openly available softwares and standards. More recently, my research activities has expanded and I developed an interest in the data science field, with emphasis in systems engineering techniques (such as machine learning algorithms and data mining) for (bio)chemical & biomedical data analysis and hybrid mechanistic/ML methods, in order to improve outcomes and better decision-making. Moreover, I am currently involving on the regular maintenance of KiMoSys, a database for quantitative kinetic models, and IPOscore, a web-based platform for assisting clinical decisions in the oncology domain.

Key-terms of research: Systems Biology modeling, optimization, industrial biotechnology, (bio)process&systems engineering, bioinformatics, metabolism, (bio)process digitalization, computational medicine&biology, systems biotechnology, data management, Bioindustry 4.0, metabolic engineering, data analytics, bioengineering

Main Topics

We are working on computational solutions to address biotechnological, (bio)chemical and biomedical problems. Our work is highly interdisciplinary (at the interface of bioengineering, computer science, biotechnology, medicine, chemistry and biology), focused at:

☛ Mathematical models of microbial metabolism for sustainable biotechnological applications ([P1], [P6], [P9], [P10]).

☛ Modeling, optimization & automation of bioprocesses and biosystems ([P3], [P4], [P8], [P11]).

☛ Computational tools and methods (machine learning and data mining) to analyse big-datasets coming from diverse sources (e.g., (bio)chemical data [P2], environmental [P7] and clinical/biomedical data [P5]).

Interested in applying computational models to biotechnological and biomedical challenges? Please Contact me

Selected Publications

[P1] J. Pinto, J. Ramos, R. S. Costa & R. Oliveira, A General Hybrid Modeling Framework for Systems Biology Applications: Combining Mechanistic Knowledge with Deep Neural Networks under the SBML Standard. AI, 4(1), 303-318, (2023), doi: 10.3390/ai4010014

[P2] Alexandre, L., Costa, R.S. & Henriques, R. DI2: prior-free and multi-item discretization of biological data and its applications. BMC Bioinformatics, 22, 426 (2021), doi: 10.1186/s12859-021-04329-8, CODE

[P3] Gonçalves, D., Henriques, R., Costa, R.S. Predicting metabolic fluxes from omics data via machine learning: Moving from knowledge-driven towards data-driven approaches. Computational and Structural Biotechnology Journal, 21, 4960-4973, (2023), doi: 10.1016/j.csbj.2023.10.002, CODE

[P4] J. Pinto, J. Ramos, R. S. Costa & R. Oliveira. Hybrid Deep Modeling of a GS115 (Mut+) Pichia pastoris Culture with State–Space Reduction. Fermentation, 9(7), 643, (2023), doi: 10.3390/fermentation9070643

[P5] Alexandre, L., Costa, R.S., Santos, L. L., Henriques, R. Mining pre-surgical patterns able to discriminate post-surgical outcomes in the oncological domain. IEEE Journal of Biomedical and Health Informatics, 25(7), (2021), doi: 10.1109/JBHI.2021.3064786

[P6] Mochao, H., Barahona, P., Costa, R.S. KiMoSys 2.0: an upgraded database for submitting, storing and accessing experimental data for kinetic modeling. Database-The journal of biological database & curation, baaa093, (2020), doi: 10.1093/database/baaa093

[P7] Patrício, A., Lopes, M. B., Costa, P.R., Costa, R.S., Henriques, R., Vinga S. Time-Lagged Correlation Analysis of Shellfish Toxicity Reveals Predictive Links to Adjacent Areas, Species, and Environmental Conditions. Toxins, 14 (10), 679, (2022), doi: 10.3390/toxins14100679, CODE

[P8] J. Pinto, M. Mestre, J. Ramos, R. S. Costa, G. Striedner, R. Oliveira A general deep hybrid model for bioreactor systems: combining First Principles with deep neural networks. Computers & Chemical Engineering. 165, 107952, (2022), doi: 10.1016/j.compchemeng.2022.107952

[P9] Costa, R. S., Hartmann, A., Gaspar, P., Neves, A. R. and Vinga, S. An extended dynamic model of Lactococcus lactis metabolism for mannitol and 2,3-butanediol production. Molecular BioSystems 10, 628-639 (2014), doi: C3MB70265K, SBML MODEL

[P10] Machado, D., Costa, R.S., Rocha, I., Tidor B., Ferreira, E.C. Exploring the gap between dynamic and constraint-based models of metabolism. Metabolic Engineering 14 (2), 112-119 (2012), doi: 10.1016/j.ymben.2012.01.003 (Journal Cover Paper)

[P11] J. Pinto †, R. S. Costa* †, L. Alexandre, J. Ramos, R. Oliveira. SBML2HYB: a Python interface for SBML compatible hybrid modeling. Bioinformatics. 39(1), btad044, (2023), doi: bioinformatics/btad044, CODE

Full list of publications: Google Scholar - ORCID - ResearcherID

Representative Research Projects

• BioLaMer – Proof of principle fly larvae biorefinery for biopolymer plastic production (HORIZON-EIC), Coordinator: Sibu Padmanabhan. (ongoing)

• IPOscore – Predicting the risk of complications of surgical treatment and define prognosis of cancer patients through clinical and biopathological data integration, Data Science and Artificial Intelligence in Public Administration project grant, (FCT). Role: Principal Investigator.

• PROMICON – Harnessing the power of Nature through productive microbial consortia in biotechnology – measure, model, master (H2020-FNR-12-2020), Coordinator: Jens Kromer. (ongoing)

• BacHBerry – BACterial Hosts for production of Bioactive phenolics from bERRY fruits (FP7-KBBE.2013.3.1-01), Role: Team member, PI: Jochen Forster.

Software & Tools

► DISA, a software package in Python capable of assessing patterns with numerical outputs by statistically testing the correlation gain of the subspace against the overall space. To illustrate the DISA properties, a dataset that monitors the concentration of key enzymes observed in two Design-Build-Test-Learn (DBTL) cycles of 1-dodecanol production in E. coli is presented.

► score4Covid, a clinical decision support system for Covid-19 to anticipate medical needs and infection outcomes using demographic and comorbidity variables, as well as onset date of symptoms, test and hospitalization.

► DI2 tool, a python package that uses non-parametric tests to find the best fitting distribution for a given variable and discretize it accordingly. To illustrate the DI2 tool properties two case studies were analyzed including a yeast experimental dataset.

► KiMoSys database. An experimental data repository for Kinetic Models of biological SYStems. Available at kimosys.org

► ObjComparison. Matlab implementation for investigation the effect of cellular objective function and constraints in flux balance constraint-based models. Implementation of all objective functions and constraints can be adapted to test different metabolic systems. Available at http://bit.ly/17QACEW

► SON-EM. An algorithm for hybrid system identification of Switched affine AutoRegressive model with eXogenous inputs (SARX) models using Sum of Norms regularization and Expectation Maximization. Available at http://bit.ly/1DaGaVS

► SBML2HYB, a Python tool for SBML compatible hybrid modelling of biological systems. Two case studies illustrate the use of SBML2HYB: threonine synthesis pathway model and Park and Ramirez model.

► TriSig, a method designed to evaluate the statistical significance of patterns in tensor data. To demonstrate the properties of TriSig, a industrial penicillin batches dataset and a metabolomics dataset from urinary samples analyzed by NMR spectroscop are used.

Group Members

PhD Thesis IN PROGRESS
-Leonardo Alexandre ID 94054, Co-supervised with Rui Henriques (INESC-ID/IST).
-Daniel Gonçalves ID 194007, Co-supervised with Rui Henriques (INESC-ID/IST) and José A. Ferreira (IPO-Porto).
-André Patrício ID 187631, Co-supervised with Rui Henriques (INESC-ID/IST).
-Samuel Barbosa ID, Co-supervised with Lúcio L. Santos (IPO-Porto) and Daniel M. Gonçalves.

MSc Thesis IN PROGRESS
-Tomás Afonso Dias ID 84558, Risk Predicting of Surgery Complications in Cancer Patients using Deep Neural Networks with Preoperative Variables. Co-supervised with Duarte Valério (IDMEC-IST).
-André Matos ID 55358, Desenvolvimento de uma base dados da estrutura espacial em Biofilmes. Co-supervised with Pedro Barahona (FCT-NOVA) and Nuno Azevedo (LEPABE-FEUP).
-José Pedreira ID 61835 , HYBpy: a Python tool for hybrid modeling of bioprocesses. Co-supervised with Pedro Barahona (FCT-NOVA) and Rui Oliveira (FCT-NOVA).
-Marta Sousa ID 93303 , Voice disease diagnostic via machine learning. Co-supervised with Duarte Valério (IDMEC-IST).
-Maria do Carmo Ribeiro ID 96436, Predicting cardiomyocytes content using data-driven approaches.