New optical technologies enable us to see the brain in action.
Today’s sophisticated neural recording technologies enable researchers to gather data in quantities unimagined ten years ago. The challenge is to make sense of it.
New genomic and tracing technologies show the brain's intricate structure.
We are using biologically informed statistics to gain insight into psychiatric conditions, and to explore how the brain is built.
Models try to represent key aspects of brain function, and to understand how and why they change under various conditions.
Models today must integrate the wealth of new data. We draw on cell biology and computation theory.
New imaging technologies reveal fine details of behavior.
We are studying the dynamics of behavior and how brain dynamics change with aspects of behavior.
The Reimers lab does quantitative computational work in four areas related to brain dynamics and function.
- Analysis of Functional Neuroscience Data
The big data revolution has come to neuroscience. We develop methods to pre-process images and interpret the activity measures.
- Neurogenomics and Psychiatric Genetics
Genes specify the cell types and their dynamics. We develop biologically informed methods to analyze and interpret such data.
- Modeling of brain activity
We develop neural network models to match the characteristics of neural activity. We also work on genetic systems to construct neural networks that can orchestrate useful behavior.
- Analysis of high-throughput behavior data
We develop methods to characterize the fluid changes during freely behavior of animals.
Are you a mathematician, statistician, or computer scientist? Contact us to dive into some of the problems in the most exciting area of science today.
At Michigan State University I teach four courses (not every year).
I also teach Neural Data Science at Cold Spring Harbor Labs in the summers.
NEU 415 Neuroinformatics and Quantitative Reasoning
Intended audience: NEU grad students and senior undergraduates
Neuroscience students are entering a world of big data and subtle statistical issues for complex neural systems. This course will introduce them to the tools and concepts needed to function in modern neuroscience research.
There are three major components:
- Introduction to Programming using MATLAB (4 weeks)
- Advanced statistical methods and issues
Issues in robust experimental design, multiple comparisons, multivariate statistical models, permutation testing, etc. Emphasis on reproducibility.
- Using neuroscience databases
Accessing and interpreting data from the new high-throughput databases of gene expression, morphology, and connectivity.
NEU 425 Computational models in neuroscience
Intended audience: MTH, STT, CMSE and NEU senior undergraduates
Network models are often used to simulate ideas about brain function.
The aims of this course are to introduce modeling techniques and issues in the context of neuroscience theories and unresolved controversies. We will discuss successful models, practice building models, and develop critical thinking about computational models. We will study and experiment with simple models of single cells, small cortical networks, and simulations of larger network. We will study some models of simple animal behaviors and brain rhythms, and models for mammalian cognition. The course will not discuss in detail the engineering applications of computational networks for machine learning. Students will work in MATLAB and will make their own model of a specific dynamic process or behavior for a course project.
NEU 430 Genomics of brain and behavior
Intended audience: NEU and BMS senior undergraduates
Genes specify the components of the brain and how and when these are expressed. We will begin with a brief review of gene structure and regulation. Then we'll study the various gene families that code for the major distinctive components of brain cells: channels, receptors and synapses; we'll examine how and when these families have evolved, particularly in the human lineage. We will then consider how these genes vary across human populations. We will review and discuss key findings from the psychiatric and behavioral genetics literature, discuss the difficulties of doing behavioral genetics and consider why effect sizes so far associated with specific gene variants have been so small. We will discuss how experiences may cause epigenetic changes, and how such changes may affect neurons and behavior.
NEU 445 Analysis of functional neuroscience data
Intended audience: BME, MTH, STT, CMSE and NEU grad students and senior undergraduates
Today’s technologies enable neuroscientists to gather data in quantities previously unimagined, and these technologies will transform neuroscience, just as high-throughput technologies transformed genomics in the first decade of the 21st century. This course will address issues and methods in data analysis for the new high-throughput technologies in neuroscience, with emphasis on statistical issues, such as estimation accuracy and testing. The course will cover pre-processing and artifact removal, as well as spectral analysis, dimension reduction techniques and network inference. The data types considered will be multi-unit recordings, local field potentials, and optical imaging. The workshop will proceed by seminar, demonstration and practical lab data analysis exercises supervised by the instructor.
Analysis of Functional Neuroscience Data
Big data is coming to neuroscience, as to all sciences. The human brain consists of 16 Billion neurons (in the cerebrum; many more in cerebellum). They all do something different. Now we can record from ten thousand at once. New methods are needed to analyze the flood of data. The Reimers group specializes in pre-processing data generated by optical imaging with fluorescent indicators of neural activity. We also try to answer questions about brain dynamics and how different brain regions work together.
Supported by: NSF grant 1630982
Neurogenomics and Psychiatric Genetics
We work on the analysis of gene expression data in human and mouse brains, especially single-cell data, and we integrate We also works on the genetics of psychiatric disorders using chromatin information and Bayesian statistics to identify many more relevant loci.
Supported by: Templeton Foundation Human Agency Project
Modeling of brain activity
We develop neural network models intended to match the statistics of neural activity obtained under area 1. We also work on genetic models inspired by work done in area 2 to construct simulated neural networks that drive software robots.
Analysis of high-throughput behavior data
Animal (or human) behavior has seemingly infinite variability. How can we identify automatically the key movements that may be key to understanding the animal's state of mind or what will come next. We develop methods to understand fluid changes of behavior of freely behaving animals.