Thesis (Ph. D.)--Massachusetts Institute of Technology, Computational and Systems Biology Program, 2010. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references. / Excessive collagen degradation (collagenolysis) has been implicated in a series of diseases such as tumor metastasis, atherosclerosis and arthritis. There are still several unresolved questions about the mechanism of collagenolysis. First, the prototypical structure of the collagen triple helix does not fit into the active site of collagenases, the enzymes responsible of cleaving collagen. Moreover, the scissile bond that is degraded during collagenolysis is hidden from solvent. Therefore it is widely agreed that collagen unfolding must occur in order for collagenolysis to proceed. Some proposed mechanisms suggest that collagenases actively unfold collagen in order to expose the cleavage site, but no direct evidence of such mechanisms has been provided. Second, while several potential cleavage sites exist in the sequence of collagen, only one is cleaved in triple helical collagen. The hypothesis of this work is that locally unfolded states exist in collagen in the absence of collagenases. They occur as a result of the natural thermal fluctuations in the structure of collagen. Collagenolysis occurs when collagenases bind and cleave these unfolded states. In this work, a combination of computational and experimental methods is presented in order to test this hypothesis. Initially, computational results suggest that locally unfolded states are ubiquitous along the structure of collagen. However, it is shown that not all unfolded states are created equal, and that the precise sequence in the vicinity of the true collagenase cleavage site in type III collagen allows collagen to sample locally unfolded states that are complimentary to the collagenase active site. Therefore, it is hypothesized that cleavage site specificity is encoded in the nature of the unfolded states. Next, it is shown that types I and III collagen can be bound and cleaved at the actual cleavage site by just the catalytic domain of collagenases, a finding in apparent contradiction with previous work in this field. These results are interpreted in light of a novel conformational selection mechanism in which collagenases only cleave locally unfolded, vulnerable states. Finally, based on the new mechanism of collagenolysis presented here, new strategies to regulate collagenolysis are proposed. / by Ramon Salsas Escat. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Computational and Systems Biology Program, 2016. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references (pages 163-186). / Over the past decade, the advent of next-generation DNA sequencing technologies has ushered in an exciting era of biological research. Through large-scale sequencing projects, scientists have begun to unveil the variability and function of millions of DNA mutations called single nucleotide polymorphisms. Despite this rapid growth in understanding, short tandem repeats (STRs), genomic elements consisting of a repeating pattern of 2-6 bases, have remained poorly understood. Mutating orders of magnitude more rapidly than most of the human genome, STRs have been identified as the causal variants in diseases such as Fragile X syndrome and Huntington's disease. However, in spite of their potentially profound biological consequences, STRs remain systematically understudied due to difficulties associated with obtaining accurate genotypes. To address this issue, we developed a series of bioinformatics approaches and applied them to population-scale whole-genome sequencing data sets. Using data from the 1000 Genomes Project, we performed the first genome-wide characterization of STR variability by analyzing over 700,000 loci in more than 1000 individuals. Next, we integrated these genotypes with expression data to assess the contribution of STRs to gene expression in humans, uncovering their substantial regulatory role. We then developed a state-of-the-art algorithm to genotype STRs, resulting in vastly improved accuracy and uncovering hundreds of replicable de novo mutations in a deeply sequenced trio. Lastly, we developed a novel approach to estimate mutation rates for STRs on the Y-chromosome (Y-STR), resulting in rates for hundreds of previously uncharacterized markers. Collectively, these analyses highlight the extreme variability of STRs and provide a framework for incorporating them into future studies. / by Thomas F. Willems. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Computational and Systems Biology Program, 2013. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references (p. 112-119). / Microbes play a central role in diverse processes ranging from global elemental cycles to human digestion. Understanding these complex processes requires a rm under- standing of the interplay between microbes and their environment. In this thesis, we utilize sequencing data to study how individual species adapt to different niches, and how species assemble to form communities. First, we study the potential temperature and salinity range of 16 marine Vibrio strains. We nd that salinity tolerance is at odds with the strains' natural habitats, and provide evidence that this incongruence may be explained by a molecular coupling between salinity and temperature tolerance. Next, we investigate the genetic basis of bacterial ecological differentiation by analyzing the genomes of two closely related, yet ecologically distinct populations of Vibrio splendidus. We nd that most loci recombine freely across habitats, and that ecological differentiation is likely driven by a small number of habitat-specic alle-les. We further present a model for bacterial sympatric speciation. Our simulations demonstrate that a small number of adaptive loci facilitates speciation, due to the op- posing roles horizontal gene transfer (HGT) plays throughout the speciation process: HGT initially promotes speciation by bringing together multiple adaptive alleles, but later hinders it by mixing alleles across habitats. Finally, we introduce two tools for analyzing genomic survey data: SparCC, which infers correlations between taxa from relative abundance data; and StrainFinder, which extracts strain-level information from metagenomic data. Employing these tools, we infer a rich ecological network connecting hundreds of interacting species across 18 sites on the human body, and show that 16S-defined groups are rarely composed of a single dominant strain. / by Jonathan Friedman. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Computational and Systems Biology Program, February 2013. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. "September 2012." / Includes bibliographical references. / All cells sense and respond to their environments using signal transduction pathways. These pathways control a sweeping variety of cellular processes across the domains of life, but the pathways are often built from a small, shared set of protein domains. At the core of tens of thousands of signal transduction networks in bacteria is a pair of proteins, a histidine kinase and a response regulator. Upon receiving an input signal, a histidine kinase autophosphorylates and then catalyzes transfer of its phosphoryl group to a cognate response regulator, which often activates a transcriptional response. Bacteria typically encode dozens of kinases and regulators, and the kinases function as dimers in all known examples. This dimeric state raises two functional questions. Do histidine kinases specifically form dimers? Once a kinase has dimerized, does a chain in the dimer phosphorylate itself (cis) or its partner chain (trans)? Specific kinase dimerization is likely important to avoid detrimental crosstalk between separate signaling pathways, and how autophosphorylation occurs is central to kinase activity. In my thesis, I have taken biochemical and evolutionary approaches to identify molecular determinants for both dimerization specificity and autophosphorylation. To study dimerization specificity, I developed an in vitro binding assay to measure kinase dimerization, and I then showed that a paralogous pair of kinases from E. coli specifically formed homodimers over heterodimers. Residues important for dimerization specificity were predicted by measuring amino acid coevolution within kinases, which leverages the enormous amount of sequence information available for the kinase family. Experimental verification of these predictions showed that a set of residues at the base of the kinase dimerization domain was sufficient to establish homospecificity. This same region of the kinase, in particular the loops at the base of the kinase dimer, was also important for determining autophosphorylation mechanism. Recent work showed that kinases could autophosphorylate either in cis or in trans, and I found that a trans kinase could be made to autophosphorylate in cis by replacing its loop with the loop from a cis kinase. I also found that two sets of orthologs, despite having significantly diverged loop sequences, had conserved their autophosphorylation mechanisms. This raised the possibility that kinase loops may be under selection to maintain the same autophosphorylation mechanism. / by Orr Ashenberg. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Computational and Systems Biology Program, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 82-85). / Whether at the level of a single protein, or in the cytoplasm as a whole, the diffusive mobility of proteins plays a key role in biological function. To measure protein diffusion in cells, researchers have developed multiple fluorescence microscopy methods, and have tested them rigorously. However, using these methods for precise measurement of diffusion coefficients requires expertise that can be a barrier to broad utilization of these methods. Here, we report on a new method we have developed, which we name Photo-converted Intensity Profile Expansion (PIPE). It is a simple and intuitive technique that works on commercial imaging systems and requires little expertise. PIPE works by pulsing photo-convertible fluorescent proteins, generating a peaked fluorescence signal at the pulsed region, and analyzing the spatial expansion of the signal as diffusion spreads it out. The width of the expanding signal is directly related to the protein ensemble mean-square displacement, from which the diffusion coefficient of the ensemble is calculated. In the main part of the thesis, we demonstrate the success of PIPE in measuring accurate diffusion coefficients in silico, in vitro and in vivo. We then broaden the discussion, and challenge the assumption that the Fickian diffusion equation is the most appropriate model for describing protein motion in the cytoplasm. Since the cytoplasm is crowded with obstacles that trap proteins for a wide range of times, the motion of those proteins may be more accurately described by models of anomalous diffusion. To contribute to the ongoing debate about anomalous diffusion, we show how PIPE can be used to measure the degree of diffusion anomality by examining the temporal scaling of the mean-square displacement. Whether for measuring normal or anomalous diffusion, we suggest that the simplicity and user-friendliness of PIPE could make it a useful tool in molecular and cell biology. / by Rotem Gura Sadovsky. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Computational and Systems Biology Program, 2009. / Cataloged from PDF version of thesis. / Includes bibliographical references (p. 127-142). / Diversity within a population of organisms is typically ascribed to genetic differences. However, even members of a genetically identical group of cells or organisms in identical environments can exhibit variability in state and phenotype. One striking example of such heterogeneity is revealed when a genetically identical population of human cells is exposed to saturating doses of a death-inducing drug called TRAIL - many cells in the population will undergo apoptosis, a form of controlled cell death, but a fraction of cells always survives the treatment. The goal of this thesis was to understand the origins of variability in both the timing and the probability of death in TRAIL-induced apoptosis. To this end, both experimental and computational methods were implemented. Experiments examining the response of sister cells to TRAIL provided strong evidence that variability in initial conditions played a key role, and ruled out genetic, stochastic, and cell cycle effects as possible causes of heterogeneity in response. A detailed analysis of the relative contributions of three segments of the TRAIL pathway revealed that the majority of the variability in time-to-death arose upstream of mitochondrial outer membrane permeabilization (MOMP), with little contribution from downstream reactions. More specifically, the rate of cleavage of initiator caspase substrates was highly predictive of a cell's death time. However, to determine whether (as opposed to when) a cell will die, variation in the MOMP threshold became critical. / (cont.) This dependency was indicated by observation of the height of the MOMP threshold in surviving and dying cells and by modulation of this threshold via overexpression of anti-apoptotic regulators of MOMP. Simulations of cell-to-cell variability in TRAIL-induced apoptosis confirmed that the endogenous variability in apoptotic regulators was sufficient to produce the observed variability in death time. However, knowledge of the concentration of individual proteins did not allow prediction of death time because variation in other proteins masked the underlying trends. The ability to simulate heterogeneity in cellular response also led to the development of novel, biologically intuitive methods of sensitivity analysis, which revealed that sensitivities shift depending on whether knowledge of covariance in initial conditions is included. The ability to predict sensitivity and resistance of tumors to TRAIL would be clinically valuable, as TRAIL is currently in clinical trials as an anti-cancer therapy. The results described here represent progress toward understanding the "fractional killing" of tumor cells following exposure to chemotherapy, and for understanding variability in mammalian signaling pathways in general. / by Sabrina Leigh Spencer. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Computational and Systems Biology Program, 2015. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 123-138). / Humans can quickly and effortlessly recognize objects, and people and their actions from complex visual inputs. Despite the ease with which the human brain solves this problem, the underlying computational steps have remained enigmatic. What makes object and action recognition challenging are identity-preserving transformations that alter the visual appearance of objects and actions, such as changes in scale, position, and viewpoint. The majority of visual neuroscience studies examining visual recognition either use physiology recordings, which provide high spatiotemporal resolution data with limited brain coverage, or functional MRI, which provides high spatial resolution data from across the brain with limited temporal resolution. High temporal resolution data from across the brain is needed to break down and understand the computational steps underlying invariant visual recognition. In this thesis I use magenetoencephalography, machine learning, and computational modeling to study invariant visual recognition. I show that a temporal association learning rule for learning invariance in hierarchical visual systems is very robust to manipulations and visual disputations that happen during development (Chapter 2). I next show that object recognition occurs very quickly, with invariance to size and position developing in stages beginning around 100ms after stimulus onset (Chapter 3), and that action recognition occurs on a similarly fast time scale, 200 ms after video onset, with this early representation being invariant to changes in actor and viewpoint (Chapter 4). Finally, I show that the same hierarchical feedforward model can explain both the object and action recognition timing results, putting this timing data in the broader context of computer vision systems and models of the brain. This work sheds light on the computational mechanisms underlying invariant object and action recognition in the brain and demonstrates the importance of using high temporal resolution data to understand neural computations. / by Leyla Isik. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Computational and Systems Biology Program, 2012. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references (p. 126-135). / I present three novel computational methods to address the challenge of identifying protein-DNA interactions at high spatial resolution from noisy ChIP-Seq data. I first present the genome positioning system (GPS) algorithm which predicts protein-DNA interaction events from ChIP-Seq data using a single-base resolution generative probabilistic model. Using synthetic and actual ChIP-Seq data, I show that GPS improves the effective spatial resolution and accuracy in resolving proximal binding events when comparing with existing methods. Second, I present the k-mer set motif (KSM) representation and the k-mer motif alignment and clustering (KMAC) method which discovers DNA-binding motifs from ChIP-Seq derived sequences. I demonstrate that the KSM model is more predictive than the widely used position weight matrix model, and that KMAC outperforms other existing motif discovery programs in recovering known motifs from a large collection of human ChIP-Seq experiments. Finally, I present an integrative method, genome wide event finding and motif discovery (GEM), which models ChIP data with explanatory motifs and binding events at high spatial resolution. The GEM model links binding event discovery and motif discovery with positional priors in the context of a generative probabilistic model of ChIP data and genome sequence. I show that GEM further improve upon previous methods for processing ChIP-Seq and ChIP-exo data to yield unsurpassed spatial resolution and discovery of proximal binding events. GEM enables a systematic analysis of in vivo transcription factor binding to discover hundreds of spatial binding constraints between factors in human and mouse cells, including known factor pairs and novel pairs such as c-Fos:c-Jun/USF1, CTCF/Egr1, and HNF4a/FOXA1. I also discovered a complex spatial binding relationship involved 6 key regulatory factors in mouse embryonic stem (ES) cell that is likely to be functional in ES cell gene regulation. Such computational discoveries propose testable models for regulatory factor interactions that will help elucidate genome function and the implementation of combinatorial control. / by Yuchun Guo. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Computational and Systems Biology Program, 2014. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Signal transduction pathways rely on transient yet specific protein-protein interactions. How a limited set of amino acids can enforce cognate protein interactions while excluding undesired pairings remains poorly understood, even in cases where the contacting residues have been identified on both protein partners. To tackle this challenge, I performed structure-guided and library-based mutagenesis studies of bacterial two-component signaling pathways. These pathways, typically consisting of a histidine kinase and a response regulator, are an ideal model system for studying protein-protein interactions as they rely almost exclusively on molecular recognition for specificity. The kinase uses a limited set of residues to recognize the regulator in both phosphorylation and dephosphorylation reactions, and to prevent docking with all noncognate regulators. In this thesis I characterized the extent to which interface residues in two-component signaling proteins can be modified without changing the overall behavior of the pathway. In collaboration with another research group I have performed a mutagenesis study of a two-component system from Thermotoga maritima that has proven amenable to structural analysis. By solving the cocrystal structure of a histidine kinase and response regulator containing interface residues from a different interacting pair, we learned the biophysical basis for accommodating these new residues. To understand how many different residue combinations can support a functional interaction, I comprehensively mapped the sequence space of the interface formed by Escherichia coli histidine kinase PhoQ and its partner PhoP. I used a robust high-throughput assay to screen a library of 204 (160,000) PhoQ variants in which I had completely randomized the four key specificity-determining residues. Using deep sequencing, I identified -1,600 (1 %) variants that can phosphorylate and dephosphorylate PhoP as well as the wild-type PhoQ. Strikingly, PhoQ can interact with PhoP via many sets of interfacial residues that are completely different from the wild type. This combinatorial approach to mapping sequence space revealed interdependencies between individual amino acids, illustrating its power relative to screens that only examine substitutions at individual sites. This thesis provides a framework for mapping the sequence space of histidine kinases and has broad implications for understanding protein-protein interaction specificity and the evolution of bacterial signaling pathways. / by Anna Igorevna Podgornaia. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Computational and Systems Biology Program, 2010. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references (p. 117-126). / Present day microbial genomes are the handiwork of over 3 billion years of evolution. Comparisons between these genomes enable stepping backwards through past evolutionary events, and can be formalized using binary tree models known as phylogenies. In this thesis, I present three new phylogenetic methods for gaining insight into how microbes evolve. In Chapter 1, I introduce the algorithm AdaptML, which uses strain ecology information to identify genetically- and ecologically-distinct bacterial populations. Analysis of 1000 marine Vibrionaceae strains by AdaptML finds evidence that niche adaptation may influence patterns of genetic differentiation in bacteria. In Chapter 2, I introduce the algorithm AnGST, which can infer the evolutionary history of a gene family in a chronological context. Analysis of 3968 gene families drawn from 100 modern day organisms with AnGST reveals genomic evidence for a massive expansion in microbial genetic diversity during the Archean eon and the gradual oxygenation of the biosphere over the past 3 billion years. Lastly, I introduce in Chapter 3 the algorithm GAnG, which can construct prokaryotic species trees from thousands of distinct gene trees. GAnG analysis of archaeal gene trees supports hypotheses that the Nanoarchaeota diverged from the last ancestor of the Archaea prior to the Crenarchaeota/Euryarchaeota split. / by Lawrence A. David. / Ph.D.