Protein Attributes-Based Predictive Tool in a Down Syndrome Mouse Model: A Machine Learning Approach
Down syndrome is a disorder caused by an imbalance in the 21 chromosome, affecting learning and memorizing abilities, which was shown to be improved with memantine administration. In this study we intent to determine the most relevant proteins that could play a role in learning ability, suitable for possible biomarkers and to evaluate the accuracy of several bioinformatic models as a predictive tool. Five different supervised learning models (K-NN, DT, SVM, NB, NN) were applied to the original database and the newly created ones from eight attribute weighting models. Model accuracies were calculated through cross validation. Nine proteins revealed to be strong candidates as future biomarkers and K-NN and Neural Network had the better overall performances and highest accuracies (86.26% ± 0.23%; 81.51% ± 0.48%), which makes them a promising predictive tool to study protein profiles in DS patients’ follow-up after treatment with memantine.
KeywordsDown syndrome Prediction Learning improvement Attributes weighting Data mining Classification
This study was funded by QVida+: Estimação Contínua de Qualidade de Vida para Auxílio Eficaz à Decisão Clínica, NORTE‐01‐0247‐FEDER‐003446, supported by Norte Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, through the European Regional Development Fund (ERDF) and strategic project LIACC (PEst-UID/CEC/00027/2013).
- 2.Kazemi, M., Salehi, M., Kheirolahi, M.: Down syndrome: current status, challenges and future perspectives. Int. J. Mol. Cell. Med. (IJMCM) 5(3), 125–133 (2016)Google Scholar
- 7.Feng, B., Hoskins, W., Zhou, J., Xu, X., Tang, J.: Using supervised machine learning algorithms to screen Down syndrome and identify the critical protein factors. In: Xhafa, F., Patnaik, S., Zomaya, A. (eds.) Advances in Intelligent Systems and Interactive Applications, vol. 686, pp. 302–308. Springer, Cham (2018)CrossRefGoogle Scholar
- 9.Zhao, Q., Rosenbaum, K., Okada, K., Zand, D.J., Sze, R., Summar, M., Linguraru, M.G.: Automated Down syndrome detection using facial photographs. In: 2013 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (2013)Google Scholar
- 14.Victorino, D.B., Bederman, I.R., Costa, A.C.S.: Pharmacokinetic properties of memantine after a single intraperitoneal administration and multiple oral doses in euploid mice and in the Ts65Dn mouse model of Down’s syndrome. Basic Clin. Pharmacol. Toxicol. 121(5), 382–389 (2017)CrossRefGoogle Scholar
- 15.Herault, Y., Delabar, J.M., Fisher, E.M.C., Tybulewicz, V.L.J., Yu, E., Brault, V.: Rodent models in Down syndrome research: impact and future opportunities, 1 October 2017Google Scholar
- 18.Down Syndrome Memantine Follow-up Study - Full Text View - ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT02304302
- 19.Higuera, C., Gardiner, K.J., Cios, K.J.: Mice protein expression data set. UCI MLRep (2015)Google Scholar
- 23.Bustos, F.J., Jury, N., Martinez, P., Ampuero, E., Campos, M., Abarzua, S., Jaramillo, K., Ibing, S., Mardones, M.D., Haensgen, H., Kzhyshkowska, J., Tevy, M.F., Neve, R., Sanhueza, M., Varela-Nallar, L., Montecino, M., van Zundert, B.: NMDA receptor subunit composition controls dendritogenesis of hippocampal neurons through CAMKII, CREB-P, and H3K27ac. J. Cell. Physiol. 232(12), 3677–3692 (2017)CrossRefGoogle Scholar