Hulda

Thesis Title: The systematic approach in protein interaction during forgetting 

Abstract:

Active forgetting is switched on to protect our brain from overloading when get exposed to new memory contexts during memory retention. It involves multiple levels of interacting protein pathways and operates as a temporal function. The biological regulation of systems, such as electrophysiology, molecular pathways or cellular ensembles, typically proceeds in a temporally graded that allows the past events to modify ongoing behavior. Since memory is dynamic and multi-levels, with each level represented as a distributed combination of distinct entities forming a hierarchy of cause and effect continuously, we still need further research to comprehend this complex mechanism. 

Understanding of interplay between functional proteins within various time windows is important for studying memory process. The timing of postsynaptic and protein events is converted into a hierarchy of homeostatic perturbations that influence the effect of future events, therefore altering behavior. Since we considered this adaptive, temporally regulated integration of neural system memory, we dissect the memory into computable elements for analysis. In this study we inspect memory paradigms into operational components with time series, experimental manipulations and target proteins. Two models were constructed to clarify the relations between protein interaction map and animal behavior. 

There are 3 stages in experiments. In the first stages we extract information from protein interaction databases to construct operational rules possessed Markov property based on the small G protein’s pathway domain. Using published data from 2000-2017 as input, we calculate the protein relative abundance of a series of time points. In the second stage, we used the calculated protein relative abundance as training set to train a joint neural network models to predict animal behavior in real life. Finally, for evaluation of models, we used data published in 2018 as the testing set and the model’s accuracy is more than 70%. 

This study linked the regulations of protein interaction level with animal behavior over a continuous time and sheds light on future study of molecular mechanisms towards forgetting process.

Keywords: Active Forgetting, Markov Chain, Small G protein, Protein Interaction, Deep Learning

2019 Research Grant (6.5k€), Ministry of Science and Technology, China.

2016-2019 Scholarship (~10k€), Ministry of Education Fellowship. China.

Associate Project:

“Reduced Smoothened level rescues A-beta-induced memory deficits and neuronal inflammation in animal models of Alzheimer's disease”

“Long-term memory is formed immediately without the need for protein synthesis-dependent consolidation in Drosophila”

Skills developed: Animal Experiment (Drosophila and mice), Neural Circuit, Cognitive Neuroscience, Neural Computation.