Projects

Gunawardena Lab

The Gunawardena Lab studies cellular information processing using a combination of mathematical and experimental techniques. In recent years, the lab has shifted from being primarily experimental to being primarily theoretical, in part because our experiments are now done in collaboration and in part because the theoretical problems have become more challenging. We have a tradition of mentoring undergraduate students into scientific research and welcome students from groups who are under-represented in science. Several of our students have been first authors on papers arising from their undergraduate work (http://vcp.med.harvard.edu/people). Students come from a range of backgrounds and usually work on mathematical projects arising from the main research directions in the lab. At present, we are interested in the following kinds of questions, which can be explored further through our papers (http://vcp.med.harvard.edu/papers.html). 

(1) How is information encoded by protein post-translational modification?

(2) How does energy expenditure allow cells to process information better?

(3) How do biochemical mechanisms function robustly while being plastic on

evolutionary timescales?

Our lab offers a halfway house between the biological and the mathematical sciences. If you are not scared of mathematics, have a genuine interest in modern biology and are willing to work hard for a couple of months, you could have a lot of fun.

Megason Lab:

Computer recreation of the beginning of life” Life begins when the single cell egg is fertilized by a sperm which triggers it to divide repeatedly until there are tens to thousands (depending on the species) of seemingly equivalent cells that then go on to adopt different identities. During this early cleavage stage, cells divide in stereotypic, but not invariant, lineages. To what extent these cell lineage patterns can be explained by mechanical/geometrical influences on cell shape and division orientation vs. molecular determinants remains an open question. In this project, we will attempt to computationally simulate cell division, cell adhesion, and cell shape from a physics (but not genetic) perspective from the 1 cell to 4000 cell stage of zebrafish embryogenesis to explore to what extent physics vs genetics controls early development. We will use Python scripts of a popular open-source, 3D computer animation package called Blender to create Pixar like movies of the first stage of life. Our goal is to simulate the first 4 hours of this movie of zebrafish development- https://www.youtube.com/watch?v=EeHiA98yUa4

Pehlevan Lab:

Biological brains have helped inspire the development of artificial computers throughout history, culminating most recently in the development of artificial neural networks. This project investigates specific ways to further incorporate biologically-inspired details and constraints into artificial neural network models. The development of more brain-like models has a dual purpose of opening the potential for more powerful and efficient artificial computers as well as the development of more realistic models that can shed light on brain behavior and function. This project will take a normative approach, where biologically relevant resources are minimized alongside the maximization of performance. These resource constraints will modify the behavior of the network models in ways that the intern will work to understand analytically and through simulations. Possibilities include incorporating neurotransmitter effects to increase the power and flexibility of network models with the same number of synapses. Where appropriate, recently developed machinery for training spiking network models will be employed. The precise context for these experiments is flexible. One context that may be of interest is that of simple reinforcement learning tasks, where connections with biology are rich but where an understanding of the networks that learn these tasks is still limited. Others include models of simple sensory processing circuits such as those underlying insect olfaction. Examples of model properties that are of interest are: tuning properties of neural units, geometry of learned representations, model identifiability, few-shot learning, and bits computed per unit energy

Higgins Lab: 

Title: Modeling in vivo human blood cell population dynamics in health and disease using a combination of mechanistic ODE/PDE models and machine learning. Description: Blood cells and platelets provide the initial response to many diseases processes. These populations are tightly regulated in healthy people, and their dynamics change in different disease states, like trauma, infection, cancer, and autoimmunity. The typical human red blood cell circulates for about 100 days. Some white blood cells circulate for only a few hours while others remain viable in the periphery for years. The distribution of ages of different types of blood cells varies in response to patient age and disease and often provides a record of physiologic and pathologic adaptation to disease and environmental conditions. We are interested in estimating rates of cellular production and turnover and the resulting age distributions from data collected during routine blood counts.

We would like to understand how these inferred rates and distributions change in response to both healthy perturbations such as endurance training for a marathon and unhealthy perturbations such as COVID-19, iron deficiency anemia, trauma, sickle cell disease, and more. Depending on student interests and experience, more advanced investigations may be undertaken to assess blood cell morphology by analyzing digitized patient blood smears or variation in rheology of patient blood samples – as well as how these characteristics change over time in healthy individuals and in those with disease, possibly in response to treatment.

Some experience programming in MATLAB and related languages and with ordinary differential equations and machine learning methods is essential, as is interest in human physiology and disease and an eagerness to take initiative. Experience with immunology or partial differential equations is helpful.

Klein Lab:  

Cells in our body possess the same genetic material, but our tissues are composed of many distinct cell types that differ in their organization, composition and function. How are these cell types established, and how does evolution act on a common genetic blueprint to produce such non-genetic diversity? If we could answer these questions, could we one day engineer novel cell types, or better control the function of tissues? In this internship, you will join us in exploring these questions. We are focusing on the example of how evolution gave rise to red blood cells — circulating cells tasked with oxygen transport. While (almost) all vertebrates have red blood cells, most invertebrates do not have dedicated oxygen-carrying cells. We are working to map the events that led to the appearance of this novel cell type in vertebrates. By examining some key defining features of red blood cells, we will explore invertebrate organisms and track how and when in evolutionary history these features appeared. This project will focus on challenges in genomics and single cell RNA sequencing to analyze cells across species.

Lahav Lab:

Project: p53 is a transcription factor that is activated by cellular stresses such as DNA damage. It can induce diverse, and sometimes conflicting, transcriptional programs, such as cell cycle arrest to promote cell survival, or apoptosis to promote cell death. The choice of transcriptional programs is determined in part by the dynamics of p53 (i.e. its changes in levels over time) following cell stress. DNA damage induces oscillations that cause cell cycle arrest. The levels of p53 are regulated via degradation by the E3 ubiquitin ligase Mdm2 and its accessory factor Mdmx. We have recently found that even in the absence of cell stress, a chemical inhibitor against Mdmx induces p53 oscillations whose period varies with the inhibitor dose. This finding represents the first known instance of alterations to the period of p53 oscillations, and provides a unique opportunity to explore how p53 dynamical patterns influence downstream responses. The goals of this project are to define the precise relationship between inhibitor dose and p53 oscillations and to determine how the period of oscillations affects the choice of transcriptional programs. In this internship, you will use data on p53 levels collected under various doses of Mdmx inhibitor to determine how Mdmx inhibition influences various characteristics of p53 oscillations, such as period and amplitude. You will then model the expression levels of p53 target genes under different patterns of p53 dynamics to predict which gene expression programs are induced under different conditions of p53 expression. The results of this work will provide an important link in our understanding of how protein dynamics guide cellular outcomes.

Gaudet Lab

Evolutionary and structural analysis of the APC transporter superfamily The APC superfamily of secondary transporters are responsible for mediating the transport of countless diverse substrates across all domains of life, from transition metals to neurotransmitters. Despite high sequence divergence, structural studies have shown that they all share a similar three-dimensional structure. We investigate the mechanism of evolution of transport in one family of these transporters – the metal-transporting Nramps – but are also interested in systematic comparisons between the Nramps and other transporters sharing the same fold and thus the same common evolutionary ancestor. Our evidence suggests profound differences both in specificity and in the mechanism of conformational change across this superfamily, but how exactly these differences are encoded in the proteins’ sequences is still an open question. This project could take many directions, but in general can involve:

· Building phylogenetic trees and analyzing rates of evolution

· Comparing known crystal structures of APC transporter proteins

· Performing analysis of coevolution to identify key functional positions from sequence data

· Reading scientific papers to synthesize our knowledge about these exciting proteins

DePace Lab

The DePace lab studies fundamental principles of gene regulation during early development in Drosophila embryos and cultured cells. Among the directions of the lab, we are curious how enhancers encode information about spatially and temporal distinct gene expression, and how this information is decoded at the level of individual genes into the regulation of the transcription cycle.

In this project, we will work on quantitative image analysis of transcriptional activation in an engineered gene regulatory circuit. We will use fluorescence microscopy data of transcription reporters to ask about general principles how transcription factors modulate gene expression and/or the context dependent effects from surrounding chromatin. Depending on the interests and existing programming skills, the project can focus on the analysis of fluorescence microscopy data and what we can learn from it about transcription factor mediated gene expression, or on prototyping and implementing novel software tools for quantitative image analysis.

Our lab members come from a variety of backgrounds, unified by a strong interest in quantitative science with a passion for fluorescence microscopy. We celebrate a very interactive lab atmosphere and are looking forward to welcome a student that is curious to learn about image analysis and think deeply about what quantitative fluorescence microscopy data can tell us about how transcription is regulated.