Quantitative Biology Colloquium

Spring 2008

 

Schedule 4:15-5:15 Tuesdays

February 5

Chiu-Hsieh (Paul) Hsu

Department of Epidemiology & Biostatistics,

College of Public Health

Survival Analysis via Multiple Imputation

Keating 103

In survival analysis, censored observations can be regarded as missing event time data. We develop a direct approach, multiple imputation, which shares common ideas with the redistribute to the right algorithm to recover information for censored observations. This approach can handle both interval-censored and right-censored data. In one sample situations, we show that with a large number of imputes the imputation method will reproduce the Kaplan-Meier estimates. In interval-censored data situations, we propose a nonparametric multiple imputation scheme, NPMLE imputation. The imputation methods convert interval-censored data problems to completed data problems, which enables estimates of measures of uncertainty to be more easily obtained. In a situation with auxiliary variables, we incorporate auxiliary variables into imputation through two working Cox proportional hazards models and show that the imputation method can reduce bias due to dependent censoring and improve the efficiency. These findings of improved efficiency and reduced bias can be seen in both interval-censored and right-censored data. The methods are applied to AIDS data sets.

 

February 12

 

Ted Trouard

Departments of Biomedical Engineering and Radiology and BIO5 Institute

Noninvasive Measurement of Tissue Properties with MRI

MRB 102

Magnetic resonance imaging (MRI) is a common and useful tool in medicine and biology.  Typically, when MRI images are used by radiologist to make a clinical diagnosis, the analysis of the images is carried out qualitatively, and relies heavily on the experience of the radiologist. Although quantitative information is available in MRI, it is typically not obtained because of the increased data acquisition time required, and the ability of radiologists to make decisions without it.  However, in some circumstances, quantitative information can be very valuable in the diagnosis and evaluation of disease and in critical evaluation of therapy.  A description of quantitative parameters available from MRI experiments will be presented and research projects in which they are being used will be discussed Magnetic resonance imaging (MRI) is a common and useful tool in medicine and biology.  Typically, when MRI images are used by radiologist to make a clinical diagnosis, the analysis of the images is carried out qualitatively, and relies heavily on the experience of the radiologist. Although quantitative information is available in MRI, it is typically not obtained because of the increased data acquisition time required, and the ability of radiologists to make decisions without it.  However, in some circumstances, quantitative information can be very valuable in the diagnosis and evaluation of disease and in critical evaluation of therapy.  A description of quantitative parameters available from MRI experiments will be presented and research projects in which they are being used will be discussed

 

February 19

 

Gary Skinner     

Department of Physics

Mechanical Tension within DNA Controls Gene Transcription Initiation

MRB 102

We have investigated the effect of DNA tension upon the association and dissociation of the T7 RNA polymerase/promoter DNA Interaction. Using two optically trapped beads, we suspended a single DNA molecule containing the T7-phi13 promoter above a 3rd surface-immobilized bead that bears active molecules of T7 RNAP. The DNA was brought into contact with the surface bead, while applying a 50 Hz triangle oscillation to  the bead upstream of the promoter. Binding of T7 RNAP at the promoter was detected as the decoupling of motion of the two optically trapped beads: oscillations of the downstream bead are reduced or cease altogether upon binding. The DNA tension upon binding is directly proportional to the displacement of the downstream bead from the center of the optical trap. By altering both the mean tension in the DNA - changing trap stiffness or the mean bead-to-bead distance - we have been able to observe promoter binding across the range of force 1-12 pN. From these data we find clear evidence that the mean lifetime of the promoter/T7 RNAP complex decreases with increasing tension as indicated by an increasing dissociation rate constant, koff, from 3.5 s-1 at 1.5 pN up to 73 s-1 at >8 pN. From these observations, we propose that tension within a DNA molecule is able to regulate gene expression, at least in the case of the bacteriophage T7 enzyme. Based on the recent structural data of the T7 RNAP initiation complex we speculate on the mechanism by which force may affect the dissociation kinetics.

 

February 26

 

Bruce Walsh

Department of Ecology  and Evolutionary Biology and BIO5 Institute

Who's Your Great-great-great Granddaddy?  Estimation of Time to Most Recent Common Ancestor Given Molecular Genealogical Data

Keating 103

Molecular marker data is exceptional at determining if a biological sample (suchas blood or an offspring!) came from a specified individual.  Genealogists have become very interested in using molecular data to assess slightly more distant relationships, those within 3-20 generations.  I'll review why this is done using collections of completely linked markers (such as the Y chromosome) and discuss various statistical models for estimating the time to the most recent common ancestor given two such sequences.  We start with a simple maximum likelihood approach, discover its flaws, and then develop a fully Bayesian estimator that addresses these concerns.

 

March 4

 

Jonathan Overpeck

Institute for the Study of Planet Earth

Departments of Geosciences and  Atmospheric Sciences

Climate Change, Sea Level, and Western Drought: Dangerous Anthropogenic Interference?

Keating 103

The reality of global warming, reflected in a broad spectrum of climate system change, is now unequivocal. Moreover, human complicity in global warming has also been established beyond a reasonable doubt. With these and other advances embodied in the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4) has come a shift in scientific focus toward efforts designed to improve our understanding of what will happen in the future, and to what can be done to deal with the reality of global to regional climate change. Major efforts are now needed to develop strategies for adapting to climate change that is already in the pipeline, and also to identify climate changes that may be deemed unacceptable, and thus worthy drivers of mitigation strategies designed to reduce the rate of atmospheric greenhouse gas increases to safe levels.

 

There may be many aspects of future global climate change that will ultimately be deemed undesirable and worthy of mitigation efforts, but two major issues are already coming into focus. The first is global sea level rise coupled with increasing tropical storm intensities. There is no doubt that global sea level is rising, and little doubt that the rates of sea level rise are likely to increase. In contrast to what some biased media and individuals are saying, the IPCC AR4 did not lower estimates of future sea level rise. Indeed, the most recent estimates suggest that 1m or more of sea level rise could occur by 2100, as well as a commitment to a much larger sea level rise over subsequent centuries. The wildcard will be the future behavior of the large polar ice sheets, and there is growing evidence that the ice sheets are more vulnerable to global warming than widely thought.

 

Although coastal areas could thus be big losers in the face of continued climate change, recent climate change coupled with climate change projections indicate that the American West – including the alpine West - could be a more near-term casualty. Surface air temperatures are already rising faster than elsewhere in the coterminous United States, and will likely continue to rise steadily. These temperature increases are already causing snow to fall increasingly as rain, and also to melt earlier in the year. Thus, even in the absence of a precipitation decrease, there will be less snow-related run-off and related surface water flow. This trend is also being exacerbated in some parts of the West by human-caused increases in atmospheric dust loading. Unfortunately, nearly all state-of-the-art climate models being forced with increasing greenhouse gases (and other human-caused pollution) are also simulating a steady decline in average wintertime precipitation in the Southwest. More troubling is the fact that these simulated changes are also in accord with what has been happening in the real world – there is a growing scientific consensus that winters will become much hotter and significantly drier due to the greenhouse-gas climate forcing. On top of these trends in average condition is the likelihood that multi-year, even multi-decade, drought will also become more common.  Thus, the recent western drought – already the worst of the instrumental era – could be a harbinger of greater aridity to come, and also a significant threat to the West as we know it. Fortunately, there are solutions if we choose to act aggressively.

 

March 11

 

Vicki Chandler

Department of Plant Sciences and BIO5 Institute

Genome-wide Analyses of Tandem Repeats in Humans

Keating 103

Using the compiled human genome sequence, we systematically catalogued all tandem repeats with periods between 20 and 2,000 bp and defined two subsets whose pattern sequences were found at either single (slTRs) or multiple loci (mlTRs).  Parameters compiled for these subsets are consistent with a two step model for their evolution beginning with an initial duplication event insensitive to the unique or repetitive nature of the nascent sequence.  Subsequent amplification steps via gene conversion are infrequent and can create new alleles varying in both repeat number and internal sequence.  Tandem repeats are non-randomly distributed in the genome; both subsets are found at higher frequency at many but not all chromosome ends and internal clusters of mlTRs are also seen.  Despite the integral role of recombination in the biology of tandem repeats, recombination hotspots co-localized only with shorter microsatellites and not the longer repeats examined here.  An increased frequency of slTRs was observed near imprinted genes, consistent with a functional role, while both slTRs and mlTRs were found more frequently near genes implicated in triplet expansion diseases, suggesting a general instability of these regions. Using our collated parameters, we identified 2,230 slTRs as candidates for highly informative molecular markers.

 

March 18

 

Spring Break

March 28 – Al Scott Lecture

Julia Arciero

Program in Applied Mathematics

Theoretical model of metabolic blood flow regulation

4:00

Math 501

The ability of the circulatory system to adequately match blood supply to tissue demand implies the existence of regulatory mechanisms that communicate tissue status to blood vessels.  For example, red blood cells have been shown to respond to low tissue oxygen levels by releasing ATP. The ATP triggers a conducted response signal to travel upstream and cause arterioles to dilate so that more blood is delivered to the region of demand.  A theoretical model focusing on the role of this mechanism in blood flow regulation is presented here.  In the model, arterioles control blood flow by dilating or constricting in response to changes inmetabolism as well as to changes in pressure and wall shear stress.  The model predicts that responses to these three stimuli can account for the increase in blood flow that occurs with increased oxygen demand.

 

April 1

 

Postponed to April 29

April 8

 

Jonathan Dyhr

Program in Neuroscience

Visual Speedometers in the Honeybee Brain

Keating 103

In 1973 Karl von Frisch was awarded the Nobel Prize in Physiology for his work decoding the "language" of bees.  The language of bees, also known as the "waggle dance", is used by foraging honeybees to indicate the distance and direction of a food source to other bees in the hive.  Since then we have learned from numerous behavioral studies that a honeybee's judgment of distance

is based on a visual estimate of speed.  This estimate is relatively independent of spatial frequency, contrast and direction of motion and can even be made with only monocular input.

 

Despite our thorough understanding of the many behaviors that rely on visual speed estimation, we still lack a solid understanding of the underlying neural processes responsible for the estimate.  My research has focused on evaluating various models of visual speed estimation based on their effectiveness for replicating, in simulation, the behaviors observed in honeybees.  I then use

the results from the simulations to generate hypotheses and experiments to further refine the models. In addition, I have recently begun my own behavioral work to gather more detailed information on the properties of the honeybee speedometer..

 

April 15

 

Andrew Hausrath

Department of Biochemistry and Molecular Biophysics

Alain Goriely

Department of Mathematics

Coils, Coiled-coils, and Coiled-coiled-coils: Unraveling the Hierarchical Structure of Adiponectin

Keating 103

Adiponectin is a signalling hormone which stimulates the body's response to insulin and is of interest as a possible therapeutic for diabetes. Detailed knowledge of adiponectin's three-dimensional structure is prerequisite to a mechanistic understanding of its physiological role. However, its biochemical properties preclude the use of conventional experimental structure-determination methods. In this talk we will show how to develop new methods, based on the differential geometry of curves to model the structure and mechanics of proteins. These methods are particularly well-suited to explore the properties of helical assembly. Starting with simple helical structure (coils) we will show how we can construct higher order assembly such as the ubiquitous coiled-coils, a protein motif found in many fibrous proteins. The mechanical properties of such assembly can be determined by developing a multiscale theory. The next higher order assembly is the coiled-coiled-coil, which we believe is applicable to adiponectin, since it  exists in serum in three distinct oligomeric states with 3, 6, and 18 protein subunits.

 

April 22

 

Parker Antin

Departments of Cell Biology and Anatomy and Molecular and Cellular Biology

Leo Lopes

Department of Systems and Industrial Engineering

Network Modeling of Gene Regulatory Pathways in Vertebrate Development

Keating 103

We are developing methods for defining transcriptional regulatory networks underlying fundamental processes during embryonic development. Our general approach is to create a preliminary structure network (PSN) using a variety of biological information stores, to iteratively sample this and modified networks across parameter space, and then to test successful networks in vivo. A primary limitation for generation and modification of the PSN is the relative lack of interaction data in most model organisms, including chicken. To overcome this problem, we have created integrated interolog (interaction homolog) networks for all organisms represented in NCBIs HomoloGene database using all available interaction data. This has greatly increased the amount of interaction data available for the twenty organisms presently in HomoloGene. We have also developed new algorithms for more rapidly sampling networks in topology space and for measuring their robustness with respect to changes in reaction rates. To measure robustness efficiently, we use approximations to the ODE systems that are valid in a neighborhood, and use the fact that re-optimizing is much less expensive than optimizing from scratch. Our methods are related to Markov Chain Monte Carlo sampling and Trust Regions. Preliminary results indicate an improvement of an order of magnitude in terms of computing time, with only a marginal increase in variance.

 

April 29

 

Herman Gordon

Departments of Cell Biology and Anatomy and Molecular and Cellular  Biology

Multiple Models of Molecular Assembly at Synapses

Keating 103

Assembly of molecular specializations in cell membranes clearly involves signaling. Examples include neural synapses, immunological synapses, and cell surface adhesion structures.  But HOW does the signaling get translated into the formation of large scale molecular structures?  Also, the resulting structures must be quasi-stable, accommodating the incessant turnover of cell surface molecules while retaining characteristic sizes and substructure over potentially years.

 

I present multiple models that generate molecular assemblies with remarkable similarities to those seen in biological systems.  Successful models are self-organizing and involve coupled state transitions.  Rich structures result even from the self-organization of a single type of molecule.  The use of multiple models helps to identify specific properties that underly aspects of the biology.  For example, the perforation of synapses from central and peripheral nervous systems and even including immunological synapses can result from the selective turnover of anchored molecules within the synapses.