This is an informal overview of the job opportunities.
Please contact tetsuya (at) sat.t.u-tokyo.ac.jp for more information.
In laboratory for quantitative biology @IIS, UTokyo, we are working on understanding and prediction of various biological information processing by combining quantitative data analysis and mathematical modeling.
The successful candidate(s) of this opportunity are supposed to engage in one or two following projects:
(1) Understanding and prediction of molecular recognition mechanisms in mammalian olfaction and adaptive immunity
By receiving, recognizing and processing extremely complex and diverse chemical information from the environment, an organism selects and induce complex cellular responses and differentiation pathways (signaling system), which are prominent in odor recognition (olfactory system) and antigen recognitions (immune system). In addition, we now know that intracellular reactions and signalling pathways may be more promiscuous and less specific than what people presumed.
In this project, we aim to understand the mechanisms of information processing in molecular recognition by constructing theories and data analysis methods for predicting multidimensional input-output relationships. The olfactory and immune systems project distributed information acquired by a huge variety of receptors onto a high-dimensional space using random projections, from which chemical substances are classified. We are trying to understand the roles of the network structure specific to chemical substance recognition and to develop new methods for data analysis.
These theoretical predictions will be validated using simultaneous imaging of parallel signaling systems, comprehensive olfactory receptor response data, and sequence data on immune receptor repertoire.
We also develop a general mathematical theory for chemical reactions and chemical reactions networks by using techniques from various fields such as chemical thermodynamics, graph theory, topology & homology, information & algebraic geometry, and so on.
This research will be conducted as parts of JST CREST project and of a joint research project with Ajinomoto Co.
keywords: chemical reaction theory, chemical reaction network theory, multi-terminal information theory, compressed sensing, active sensing, computational algebra, toric variety, cohomology, chemoinformatics, graph neural network, network analysis, time-series analysis, information bottleneck, next generation sequencing, GPCR, olfaction, immuno-repertoire, sensor array.
(2) Information physics of adaptive and learning behaviors at intracellular, single-cell, and multi-cell levels.
A cell receives signals from the environment and responds to them in a variety of ways, by switching its phenotype, modulating its locomotion, and in some cases, also changes the environment itself.
This loop of sensing and response enables us to understand the adaptation and learning processes at the cellular level from the standpoint of information theory, optimal control, and learning theory. In addition, by utilizing multi-agent learning theory and game theory, we will also address adaptation and learning at the population level. On the other hand, these adaptation and learning responses are realized by various functions of intracellular chemical reactions (differential and integral operations, molecular recognition, noise and perturbation tolerance, self-replication, etc.). We will work on elucidating the structures of the networks required to implement these functions and the quantitative relationships with the thermodynamic costs that are inevitably associated with the functionality.
We aim to validate the theoretical predictions of regulatory structures and quantitative properties based on quantitative experimental measurements of chemotaxis systems in E. coli , cellular slime mold, and cell signaling systems, combined with methods such as inverse reinforcement learning.
This research will be conducted as part of "Information Physics of Life matters" project (2019/06~2023/03).
keywords: information theory, filtering theory, optimal control, reinforcement learning, computational neuroscience, Markov decision process, statistical decision theory, stochastic thermodynamics, stochastic process, inverse reinforcement learning, game theory, chemotaxis, gradient sensing, active matter, perfect adaptation, absolute concentration robustness.
(3) Deciphering cellular state and prediction of its dynamics by live-cell-omics technology.
The "state" of a cell is defined as the complicated pattern of intracellular gene expression and metabolic state. Although omics is a pre･dominant technique for comprehensive analysis of gene expression and metabolism at the cellular level, it is an invasive technique that cannot trace dynamics of the "state" of a single cell over time. We aim to establish a live-cell omics technology to predict the information of omics from the Raman measurement, which is based on the correspondence between the transcriptome and Raman measurements reported by Kobayashi and Wakamoto ( bioRxiv version ). Live cell omics refers to a technique for estimating the state of omics in a cell from non-invasive Raman measurements of cells over time. In particular, we are trying to address questions such as "Why do omics information and Raman measurements correspond to each other?" and "What is the underlying structure?" from the viewpoint of statistics, machine learning, and learning theory. We explore the biological mechanisms underlying such a correspondence by the mathematical modeling of self-replicating systems. We will apply the live-cell omics technology to predict drug resistance of cells. This research will be conducted as part of the Wakamoto CREST project(2019/Oct-2024/Mar), in collaboration with Wakamoto Lab and Miyamoto Lab.
Learn more about laboratory for quantitative biology @IIS, UTokyo from https://research.crmind.net.
Informal enquiries may be made to Tetsuya J. Kobayashi (tetsuya ### sat.t.u-tokyo.ac.jp. replace ### with @)