Skip to main content


September 10, 2019: Tami Lieberman, MIT
De novo mutations in human gut and skin microbiomes
Host: D. Zeevi
There is an enormous potential for evolution within each of our microbiomes, with billions of new mutations being created each day. In this talk, I will highlight the power of tracking within-person evolution for understanding bacterial transmission, identifying genes critical to long-term survival, and for understanding evolutionary principles. I will present examples from infectious diseases, the gut microbiome, and the skin microbiome, and speculate on how within-person evolution may impact and inform microbiome-targeted therapies.
Relevant DOIs: 10.1016/j.chom.2019.03.007, 10.1038/ng.997
September 17, 2019: Raphael Turcotte, University of Oxford
Optical microscopy with wavefront control for neuroscience
Host: J. Nirody
Optical microscopy has become an essential tool for neuroscientists. The primary reason for this success is that light enables the interrogation of living tissue in its physiological context at the length-scale of neurons and their subcellular components. Nevertheless, the quality of optical images is depth-dependent, because biological samples induce aberrations and scatter light, which ultimately limits imaging depth. In this talk, I will discuss how wavefront control (WC), an approach enabling the sculpting of light, can alleviate these issues for brain imaging in live animals for several imaging modalities. First, I will show that WC can be implemented for two-photon fluorescence imaging to improve the quality of structural imaging and the accuracy of functional measurements. Then, I will illustrate how WC is essential for achieving super-resolution imaging (structured illumination microscopy) in the living brain. Finally, I will examine how WC plays an essential role in minimally invasive micro-endoscopy with multimode optical fibers and thus enables subcellular imaging centimeters into the brain in vivo.
September 24, 2019: Zahra Fakhraai, University of Pennsylvania
Surface-mediated peptide self-assembly to modulate surface energy
Host: J. Nirody
Amyloid-forming peptides or proteins typically form amyloid fibrils through a nucleation and growth mechanism. The nucleation step typically requires a minimum concentration of peptides in the solution, thus preventing this one-dimensional crystallization process from moving forward in dilute solutions. We demonstrate that in a dilute solution, where bulk nucleation and growth is limited, providing the right surface can promote rapid self-assembly of peptides into mono-layer thick amyloid fibrils within minutes. This process is diffusion-limited and depends on the orientation and diffusion of peptides on the surface. The deposition rate of peptides, as well as their orientation and diffusion on the surface strongly depend on the strength of the peptide-surface interactions. The self-assembly can be suppressed either by reducing the interaction such that the deposition rate is reduced or by increasing the interaction energy to a point where the peptide diffusion is substantially reduced.
Once the self-assembly moves forward with time, the side-chains of the residues can modulate the surface energy and thus the interaction energy between the incoming peptides and the effective surface. As such, depending on the nature of the side chains, the surface can become more hydrophobic or hydrophilic and promote or prevent the further deposition of the peptides. We demonstrate the generality of using this method to modulate the effective surface energy by the desired amount through the design of side chains, thus engineering surfaces with self-limiting self-assemblies with specific surface energy values.
October 1, 2019: Yuhai Tu, IBM
Optimal coding strategies in the peripheral olfactory systems: Compressed sensing for an array of nonlinear olfactory receptor neurons with and without spontaneous activity
Host: J. Nirody
The peripheral olfactory systems are capable of coding sparse mixtures of a few odorants in a high dimensional space (the number of possible odorant molecules is huge) by using a relatively small number of olfactory receptor neurons. Data compression is also an important problem in computer science, where powerful algorithms have been developed for compressing sparse high-dimensional data. In particular, the compressed sensing (CS) theory has been successfully used to compress high-dimensional information efficiently by exploiting the sparsity of the signal. However, the much-celebrated CS theory/algorithm requires the sensors to be linear. For neural sensory systems such as the olfactory system, the receptor neurons (sensors) respond nonlinearly to odorant concentration and have a finite response range. Therefore, the CS algorithm does not apply to sensory systems directly, and the question on how olfactory systems compress information remains open.In this talk, we will present some recent results on how a relatively small number of nonlinear sensors each with a limited response range can optimize transmission of high dimensional sparse odor mixture information. For neurons without spontaneous activity, we found that the optimal coding matrix is sparse — only a subset of neurons respond to a given odorant with their sensitivities following a broad (such as log-normal) distribution matching the odor mixture statistics. We showed that this maximum entropy code enhances the performances of the downstream reconstruction and classification tasks. For neurons with a finite spontaneous (basal) activity, our study showed that introducing odor-evoked inhibition further enhances coding capacity and the fraction of inhibitory interactions for each neuron increases with its basal activity. Comparisons with available experiments in olfactory systems are consistent with our theory.
October 10, 2019: Georg Gerber, Harvard University
Computational Biology and the Microbiome: Discovery and Prediction for Microbe-based Therapeutics and Diagnostics
Host: D. Zeevi
The human microbiome is highly dynamic on multiple timescales, changing dramatically during development of the gut in childhood, with diet, or due to medical interventions. Understanding and being able to manipulate these dynamics is essential for the rational design of microbiome-based diagnostics and therapeutics. However, analysis of longitudinal microbiome data is hampered by a paucity of tailored and principled computational methods that address inherent challenges of these data including temporally irregular and sparse sampling, experimental noise, and complex dependency structures. I will present several novel Bayesian machine learning methods that we have developed to overcome these challenges. The first, MC-TIMME (Microbial Counts Trajectories Infinite Mixture Model Engine), is a non-parametric Bayesian model for clustering microbiome time-series data that we have applied to gain insights into the temporal response of human and animal microbiota to antibiotics, infectious, and dietary perturbations. The second, MDSINE (Microbial Dynamical Systems INference Engine), is a method for efficiently inferring dynamical systems models from microbiome time-series data and predicting temporal behaviors of the microbiota, which we have applied to developing bacteriotherapies for C. difficile infection and inflammatory bowel disease. The third, Microbiome Interpretable Temporal Rule Engine (MITRE), is a method for predicting host status from microbiome time-series data, which achieves high accuracy while maintaining interpretability by learning predictive rules over automatically inferred time-periods and phylogenetically related microbes. We are currently using MITRE to analyze data from human cohorts, specifically for developing a microbiome-based diagnostic for C. difficile recurrence.
October 17, 2019: Oskar Hallatschek, Berkeley
Microbes under pressure
Host: J. Nirody
In natural settings, microbes tend to grow in dense populations where they need to push against their surroundings to accommodate space for new cells. The associated contact forces play a critical role in a variety of population-level processes, including biofilm formation, the colonization of porous media, and the invasion of biological tissues. Although mechanical forces have been characterized at the single cell level, it remains elusive how single-cell forces combine to generate population-level patterns. I present a synthesis of theory and microbial experiments that show that contact forces generated by microbial populations can become very large due to a self-driven jamming mechanism. These forces feed back on the physiology of the cells and can strongly perturb the mechanical integrity of the environment, thereby promoting microbial invasion. Finally, I highlight that the cooperative nature of microbial force generation induces a screening effect that reduces the selection against slower growing mutant types. These results underscore that, in crowded microbial populations, collective phenomena often have a mechanical basis.
October 29, 2019: Anne Dekas, Stanford University
A single-cell view of microbial activity in the dark ocean
Host: D. Zeevi
The dark ocean is one of the largest habitats for microbial life on the planet: it covers nearly two thirds of our Earth’s surface and harbors well over half of marine microorganisms. The activity of microorganisms in the deep sea plays an essential role in biogeochemical cycling, including the production and consumption of greenhouse gases (e.g., CH4, CO2 and N2O), thereby affecting climate. The goal of my research is to understand the activity of bacteria and archaea in the dark ocean: who is doing what, how much, and what affects metabolic rates? In this talk, I will describe two lines of research, one investigating nitrogen fixation in deep-sea sediments, and one probing organic substrate utilization by pelagic marine Thaumarchaeota. Additionally, I will describe the culture-independent techniques we employ, including recent methodological advances in the use of nanoscale secondary ion mass spectrometry (nanoSIMS) to quantify anabolic activity in uncultured microorganisms on the single-cell level.
November 7, 2019: Smita Krishnaswamy, Yale University
Detection structure and patterns in big biomedical data
Host: D. Zeevi
High-throughput, high-dimensional data has become ubiquitous in the biomedical and health sciences as a result of breakthroughs in measurement technologies like single cell RNA-sequencing, as well as vast improvements in health record data collection and storage. While these large datasets containing millions of cellular or patient observations hold great potential for understanding generative state space of the data, as well as drivers of differentiation, disease and progression, they also pose new challenges in terms of noise, missing data, measurement artifacts, and the so-called “curse of dimensionality.” In this talk, I will cover a unifying theme in my research which has helped to generally tackle these problems: manifold learning and the associated manifold assumption. The manifold assumption in the data analysis context refers to the idea that while the ambient measurement space is high dimensional and noisy, that the intrinsic state space lies in lower dimensional smoothly varying patches that are locally Euclidean, called manifolds. In my work, I learn the data manifold using two types of techniques: graph signal processing and deep learning. Manifold learning provides a powerful structure for algorithmic approaches to denoise the data, visualize the data and understand progressions, clusters and other regulatory patterns, as well as correct for batch effects to unify data. I will cover several applications of this principle via specific projects including: 1) MAGIC: a manifold denoising algorithm that low-pass filters data features (like audio and video signals are denoised) on a data graph, for denoising and recovery of cellular data. 2) PHATE: a general visualization and dimensionality reduction technique technique that offers an alternative to tSNE in that it preserves local and global structures, clusters as well as progressions using an information-theoretic distance between diffusion probabilities. 3) MELD: an analysis technique for comparing two or more experiments measuring the same underlying system (i.e., cells from the same type of tissue) that produces a continuously varying likelihood score throughout the manifold to indicate whether each position in the state space is enriched in one of the specific conditions. This technique is useful for pulling out subtle differences in response between different drug treatments or experimental conditions in large datasets. 4) SAUCIE (Sparse AutoEncoders for Clustering Imputation and Embedding), our highly scalable neural network architecture that simultaneously performs denoising, batch normalization, clustering and visualization via custom regularizations on different hidden layers. Finally, I will preview ongoing work in neural network architectures for predicting dynamics and other biological tasks.
November 12, 2019: Karen Kasza, Columbia University
Building the embryo: mechanisms controlling tissue flows during development
Host: J. Nirody
During embryonic development, groups of cells reorganize into functional tissues with complex form and structure. Tissue reorganization can be rapid and dramatic, often occurring through embryo-scale flows that are mediated by the coordinated actions of cells. In Drosophila embryos, cell rearrangements in the epithelium rapidly narrow and elongate the tissue, doubling the length of the body axis in just 30 minutes. This type of tissue movement is highly conserved and can be driven by internal forces generated by cells or external forces from neighboring tissues. While much is known about the molecules involved in these cell and tissue movements, it is not yet clear how these molecules work together to coordinate cell behaviors and generate coherent movements at the tissue-scale. To gain mechanistic insight into this problem, my lab combines genetic and biophysical approaches with emerging optogenetic technologies for manipulating molecular and mechanical activities in cells with high precision. I will discuss some of our recent findings on how cell properties and interactions are regulated in the Drosophila embryo to allow (or prevent) rapid cell rearrangement and tissue flow during specific developmental events.
November 18, 2019: Erik van Nimwegen, University of Basel
Inferring gene regulatory dynamics from single-cell data
Host: E. Siggia
I plan to discuss a new method for inferring, from scRNA-seq data, transcription regulatory interactions that guide single-cell gene expression trajectories. The method combines three new ideas: First, a scRNA-seq normalization method that rigorously deconvolves sampling noise from true variations in transcription rates. Second, a Bayesian method that infers the ‘regulatory states’ of each single cell by modeling measured transcription rates in terms of genome-wide computational predictions of transcription factor binding sites. And third, a maximum entropy approach that infers an effective epigenetic landscape that guides the distribution of single cells in the space of regulatory states. Time permitting I will briefly mention studies of our lab on single-cell gene regulation in bacteria using a combination of microfluidics and time-lapse microscopy. These studies highlight how gene regulation is not only strongly coupled to fluctuations in the physiological state of cells but that, by propagation of noise through the regulatory network, expression noise and gene regulation are intimately entangled.
December 3, 2019: Tamar Schlick, NYU
Folding Genes at Nucleosome Resolution
Host: J. Nirody
Deciphering chromosome tertiary organization is essential for understanding how genetic information is replicated, transcribed, silenced, and edited to control basic life processes. Many experimental studies of chromatin using nucleosome structure determination, ultra-structural techniques, single-force extension studies, and analysis of chromosomal interactions have revealed important chromatin characteristics as a function of various internal and external conditions, such as looping, compaction, and compartmentalization. Modeling studies, anchored to high-resolution nucleosome models, have explored related questions systematically. In this talk, I will describe multiscale computational approaches for chromatin modeling at nucleosome resolution and recent mesoscale chromatin simulations that incorporate key physical parameters such as nucleosome positions, linker histone binding, and acetylation marks to ‘fold’ in silico the Hox C gene cluster. The folded gene reveals a contact hub that connects an acetylation-rich with a linker histone-rich region. Such chromatin modeling techniques open the way to other computational folding of genes and genomes. Moreover, the resulting folded system emphasizes the heterogeneity of chromatin fibers and hierarchical looping motifs, and underscores how nucleosome positions in combination with epigenetic marks and linker histone binding direct the tertiary folding of fibers and genes to perform their cellular tasks. These chromatin architecture findings have important implications on many important processes including cell differentiation, gene regulation, and disease progression.
Of possible interest:
G. Ozer, A. Luque and T. Schlick, “The Chromatin Fiber: Multiscale Problems and Approaches”, Curr. Opin. Struc. Biol. 31: 124–139 (2015).
S. Grigoryev, G. Bascom, J. M. Buckwalter, M. Schubert, C. L. Woodcock, and T. Schlick, “Hierarchical Looping of Zigzag Nucleosome Chains in Metaphase Chromosomes”, Proc. Natl. Acad. Sci. USA 113: 1238–1243 (2016).
G. Bascom, K. Sanbonmatsu, and T. Schlick, “Mesoscale Modeling Reveals Hierarchical Looping of Chromatin Near Gene Regulatory Elements”, J. Phys. Chem. B Special Issue: J. Andrew McCammon Festschrift 120: 8642–8653 (2016).
G. Bascom and T. Schlick, “Linking Chromatin Fibers to Gene Folding by Hierarchical Looping”, Biophys. J. 112: 434—445 (2017).
G. Bascom, T. Kim, and T. Schlick, “Kilobase Pair Chromatin Fiber Contacts Promoted by Living-System-Like DNA Linker Length Distributions and Nucleosome Depletion”, J. Phys. Chem. B Special Issue in Memory of Klaus Schulten 121 (15): 3882–3894 (2017).
S. S. P. Rao, S.-C Huang, B. G. St. Hilaire, J. M. Engreitz, E. M. Perez, K.-R Kieffer-Kwon, A. L. Sanborn, S. E. Johnstone, G. D. Bascom, I. D. Bochkov, X. Huang, M. S. Shamim, J. Shin, D. Turner, A. D. Omer, J. T. Robinson, T. Schlick, B. E. Bernstein, R. Casellas, E. Lander, and E. Lieberman-Aiden, “Cohesin Loss Eliminates All Loop Domains, Leading to Links Among Superenhancers and Downregulation of Nearby Genes”, Cell 171: 305–320 (2017).
G. Bascom, C. Myers, and T. Schlick, “Mesoscale Modeling Reveals Formation of an Epigenetically Driven HOXC Gene Hub”, Proc. Natl. Acad. Sci. USA 116: 4955–4962 (2019). (doi: Accompanying commentary by Michele D. Pierro, “Inner Workings of Gene Folding”, Proc. Natl. Acad. Sci. USA 116: 4774–4775 (2019).
O. Perisic, S. Portillo-Ledesma, and T. Schlick, “Sensitive Effect of Linker Histone Binding Mode and Subtype on Chromatin Condensation”, Nuc. Acids Res., doi: 10.1093/nar/gkz234 (2019).
December 10, 2019: Kelley Harris, University of Washington
Evolution of the mutation rate and spectrum in diverging human and ape populations
Host: J. Nirody
Recent studies of hominoid variation have shown that mutation rates and spectra can evolve rapidly, contradicting the fixed molecular clock model. The relative mutation rates of three-base-pair motifs differ significantly among great ape lineages, implying that multiple unknown modifiers of DNA replication fidelity have arisen and fixed on each branch of the ape phylogeny. Such mutator alleles might directly modify DNA replication or repair, or might instead act indirectly by modifying traits like reproduction or chromatin structure. Certain mechanisms of action are expected to create mutations in specific regions of the genome, meaning that the spatial distribution of lineage-specific mutations is informative about their causality. To harness this source of information, we measured mutation spectra of several functional compartments (such as late-replicating regions) whose attributes are known or suspected to affect their mutation rates. Using genetic diversity from 88 great apes, we find that most functional compartments are imprinted by localized mutational signatures but that these signatures explain very little of the mutational divergence between species. Rather, compartment-specific signatures layer with species-specific signatures to create mutational portraits that reflect both lineage and function. In particular, we identify a mutation signature enriched in endogenous retroviruses that seems to co-segregate with the experimentally-measured intensity of the hydroxymethylation of retrovirus-derived DNA. Our results suggest that cis-acting mutational modifiers are highly conserved between species and rapid mutation spectrum evolution is driven primarily by trans-acting modifiers.

Past Seminars

Click here for past seminars from the Center for Studies in Physics and Biology.

Contact Us

Center for Studies in Physics and Biology
The Rockefeller University
1230 York Avenue
New York, NY 10065

$('.class-name').on('click', (evt) => { $(this).hide(); // does not run a DOM query $('.class-name').hide() // runs a DOM query });