关于申请2012美国芝加哥大学本科生暑期研究项目的通知

Department

Molecular Genetics & Cell Biology

Description

The goal of this project is to understand the mechanisms that govern the self-assembly and constriction of the actomyosin-based contractile ring during cytokinesis. We are using early embryonic cells of the nematode worm C. elegans as a model system because they divide in a highly stereotyped way; they are directlyaccessible to high resolution microscopy and they are easily manipulated using standard physical and genetic approaches. Our basic approach will combine three elements: (1) Multicolor TIRF microscopy combined with quantitative image analysis (single molecule and fluorescent speckle particle tracking, PIV) to quantify assembly, reorganization and turnover of contractile ring components at very high spatial and temporal resolution; (2) Agent-based computer simulations to explore how observed contractile ring dynamics could emerge from known properties of actin filaments, myosin motors, and their local interactions and (3) molecular and pharmacological perturbations to test key predictions of these models.

Depending on the interests and background, a student intern would work on one or more of these approaches under the guidance of a graduate student or postdoctoral fellow.

Department

Molecular Genetics & Cell Biology

Description

The Par proteins form a highly conserved system that “partitions” the cell surface into distinct and complementary domains in many different cellular contexts. In the early C. elegans embryo, the Par proteins respond to a polarizing cue by segregating into anterior and posterior domains, and then remain segregated after the cue departs. Recent studies suggest that this involves a highly dynamic competition between “anterior” and “posterior” Par proteins involving rapid exchange of proteins between the cytoplasm and the cell surface, local diffusion at the cell surface, and cross inhibitory interactions in which anterior Par proteins promote local dissociation of posterior Par proteins and posterior Par proteins promote local dissociation of anterior Par proteins.

We have recently developed methods that allow us monitor the appearance,

movement and disappearance of Par proteins at the cell surface with single molecule resolution. The goal of this project will be to use this approach to quantify

the dynamics of exchange, diffusion and competition between anterior and posterior Par proteins, then fit these data to computational models to determine how the macroscopic dynamics of polarization emerge from molecular level dynamics and interactions.

Department

Health Studies

Description

In contemporary biomedical research, it is common to measure many different features simultaneously with the hope to detect useful features that are related to disease outcome. Such data structure exhibits “large p small n” phenomenon in the sense that the dimensionality (p) far exceeds the sample size (n).

Cao and Kosorok (2011) tackled high dimensional simultaneous testing by using t-tests. Their method is shown to be robust against the hidden Markovian structure underlying different hypotheses. An important assumption they made is that the proportion of alternative hypothesis is a fixed constant. However, as the dimension grows, it is quite plausible that the number of useful features is fixed or it grows at a slower rate, thus the proportion of alternative hypotheses goes to 0.

We propose to study the multiple testing asymptotics under extremely sparse alternative hypotheses using hidden markovian model. A simple example is that the probability of switching from null hypothesis to null hypothesis is 1 − 1/p and the probability of switching from alternative hypothesis to alternative hypothesis is 1 − 1√p. In the limit stationary distribution, the proportion of alternative is a function of the dimensionality, in our example.

Department

Health Studies

Description

With the advancement of technology, more and more data emerge with “large p small n” feature, in the sense that the dimensionality (p) way exceeds the number of samples (n). This poses challenges for statistical inference. For example, when p > n, the sample covariance matrix becomes singular while the true covariance matrix is always strictly positive definite. The reason is that with the aggregation of errors, even though each entry of the covariance matrix can be consistently estimated, the overall estimation can still be poor.

Lots of interesting ideas appeared in the literature to address this issue. The most influential ones were proposed by Bickel and Levina (2008) by banding and thresholding the variance covariance matrix. While Bai (2003) first developed inferential theory for high dimensional factor models, Fan, Fan and Lv (2008) studied variance covariance matrix estimation using factor model when p < n and recently extended it to high dimensional setting.

However, they assume that the factors are observable, which is not realistic in lots of situations. We propose to use factor models for high dimensional variance and covariance matrix estimation when the factors are unobservable and study the impact on multiple testing. In genetics applications, this multivariate approach explores the relationship between different genes and potentially can yield more refined findings with improved power.

Department

Neurobiology

Description

The cephalopod community remains in the pregenomic age, which is now a major barrier to nearly everyone's research program.

There is already transcriptomic and genomic sequence available in my laboratory and others for several cephalopods, and we expect to undertake the bioinformatics in earnest, sequencing cephalopod genomes, including data sharing.

Computational skills of the student are required.

http://ragslab.bsd.uchicago.edu/

Department

Biochemistry & Molecular Biology

Description

We are presently pursuing investigation of the structural basis of uncompetitiveinhibition by an organic vanadyl chelate against protein tyrosine phospyhorylase 1B (PTP1B) using X-ray crystallographic and steady-statekinetic methods. The chelate is(acetylacetonato)oxovanadium(IV), as demonstrated in our laboratory, is the only known uncompetitive inhibitor of this enzyme (meaning that it does not bind in the active site).

Because PTP1B is critical to the regulation of cell signaling processes in normal, diabetic, and cancer cells, characterization of the binding site could generate a unique approach for development of highly specific inhibitors for potential therapeutic applications.

Department

Physics

Description

Our project involves modeling how ice melange alter river flow through a narrowing in the river channel, and comparing the results against lab experiments, field data as well as prior results on granular flowsthrough chutes. This is a collaboration with Douglas MacAyeal(Geophysics, U. Chicago) and Jason Amundson (U. Alaska, Juneau), with potential interaction with experiments in Michael Dennin's group (U.California Irvine).

Department

Physics & Math

Description

Efi Effrati, Amy Kolan and Leo Kadanoff have been working on a project aimed at bringing an old (1975) renormalization technique up to date for the twenty-first century.

This project points to an old method of calculation which has been used to get near-critical properties of Ising models and other systems.

Good knowledge of Python, MatLab, or Mathematica is required.

http://jfi.uchicago.edu/~leop/

Department

Physics & Math

Description

In recent years, another (rather analogous) method [2] has been developed to treat Hamiltonian systems. The methodology has been developed further, with new capabilities and a treatment of new problems. The capabilities include calculations of correlation functions and conformal charges. The problems include quantum phase transitions and topological order.

Good knowledge of Python, MatLab, or Mathematica is required.

http://jfi.uchicago.edu/~leop/

Department

Physics & Math

Description

We are in the process of extending the old methods based upon the new methods. We will start with Ising phase transitions in two, three, and four dimensions and move on to quantum problems and problems involving continuous variables.

Good knowledge of Python, MatLab, or Mathematica is required.

http://jfi.uchicago.edu/~leop/