XPRIME-EM: Eliciting Expert Prior Information for Motif Exploration Using the Expectation-Maximization Algorithm Wei Zhou Department of Statistics, BYU Master of Science Understanding the possible mechanisms of gene transcription regulation is a primary challenge for current molecular biologists. Identifying transcription factor binding sites (TFBSs), also called DNA motifs, is an important step in understanding these mechanisms. Furthermore, many human diseases are attributed to mutations in TFBSs, which makes identifying those DNA motifs significant for disease treatment. Uncertainty and variations in specific nucleotides of TFBSs present difficulties for DNA motif searching. In this project, we present an algorithm, XPRIME-EM (Eliciting EXpert PRior Information for Motif Exploration using the Expectation-Maximization Algorithm), which can discover known and de novo (unknown) DNA motifs simultaneously from a collection of DNA sequences using a modified EM algorithm and describe the variation nature of DNA motifs using position specific weight matrix (PWM). XPRIME improves the efficiency of locating and describing motifs by prevent the overlap of multiple motifs, a phenomenon termed a phase shift, and generates stronger motifs by considering the correlations between nucleotides at different positions within each motif. Moreover, a Bayesian formulation of the XPRIME algorithm allows for the elicitation of prior information for motifs of interest from literature and experiments into motif searching. We are the first research team to incorporate human genome-wide nucleosome occupancy information into the PWM based DNA motif searching.

With the ever-increasing complexity of sound synthesisers, there is a growing demand for automated parameter estimation and sound space navigation techniques. Recent research in this domain has focused on the application of general-purpose evolutionary algorithms to match specific types of target sounds. However, it is difficult to establish whether success or failure of a particular match is due to the inefficiency of the optimisation engine, or the limitations of the matching synthesiser. In this paper the distinction between optimiser inefficiency and synthesiser limitations is elucidated with a contrived target test methodology that enables the performance of different optimisation techniques to be measured and compared. The methodology is applied to a Frequency Modulation synthesiser, in order to compare the performance of different Evolution Strategy-based algorithms. The algorithm producing the best results with contrived targets is then used to match a non-contrived acoustic instrument tone.

With the advent of Synthetic Biology, a field between basic science and applied engineering, new computational tools are needed to help scientists reach their goal, their design, optimizing resources. In this chapter, we present a simple and powerful method to either know the DNA specificity of a wild-type protein or design new specificities by using the protein design algorithm FoldX. The only basic requirement is having a good resolution structure of the complex. Protein-DNA interaction design may aid the development of new parts designed to be orthogonal, decoupled, and precise in its target. Further, it could help to fine-tune the systems in terms of specificity, discrimination, and binding constants. In the age of newly developed devices and invented systems, computer-aided engineering promises to be an invaluable tool.

William T. Barry: Resampling-based tests of functional categories in gene expression studies (Under the direction of Dr. Fred A. Wright and Dr. Andrew B. Nobel) DNA microarrays allow researchers to measure the coexpression of thousands of genes, and are commonly used to identify changes in expression either across experimental conditions or in association with some clinical outcome. With increasing availability of gene annotation, researchers have begun to ask global questions of functional genomics that explore the interactions of genes in cellular processes and signaling pathways. A common hypothesis test for gene categories is constructed as a post hoc analysis performed once a list of significant genes is identified, using classically derived tests for 2x2 contingency tables. We note several drawbacks to this approach including the violation of an independence assumption by the correlation in expression that exists among genes. To test gene categories in a more appropriate manner, we propose a flexible, permutation-based framework, termed SAFE (for Significance Analysis of Function and Expression). SAFE is a two-stage approach, whereby gene-specific statistics are calculated for the association between expression and the response of interest and then a global statistic is used to detect a shift within a gene category to more extreme associations. Significance is assessed by repeatedly permuting whole arrays whereby the correlation between all genes is held constant and accounted for. This permutation scheme also preserves the relatedness of categories containing overlapping genes, such that error rate estimates can iii be readily obtained for multiple dependent tests. Through a detailed survey of gene category tests and simulations based on real microarray, we demonstrate how SAFE generates appropriate Type I error rates as compared to other methods. Under a more rigorously defined null hypothesis, permutation-based tests of gene categories are shown to be conservative by inducing a special case with a maximum variance for the test statistic. A bootstrap-based approach to hypothesis testing is incorporated into the SAFE framework providing better coverage and improved power under a defined class of alternatives. Lastly, we extend the SAFE framework to consider gene categories in a probabilistic manner. This allows for a hypothesis test of co-regulation, using models of transcription factor binding sites to score for the presence of motifs in the upstream regions of genes.

While the molecular mechanisms of glucocorticoid regulation of transcription have been studied in detail, the global networks regulated by the glucocorticoid receptor (GR) remain unknown. To address this question, we performed an orthogonal analysis to identify direct targets of the GR. First, we analyzed the expression profile of mouse livers in the presence or absence of exogenous glucocorticoid, resulting in over 1,300 differentially expressed genes. We then executed genome-wide location analysis on chromatin from the same livers, identifying more than 300 promoters that are bound by the GR. Intersecting the two lists yielded 53 genes whose expression is functionally dependent upon the ligand-bound GR. Further network and sequence analysis of the functional targets enabled us to suggest interactions between the GR and other transcription factors at specific target genes. Together, our results further our understanding of the GR and its targets, and provide the basis for more targeted glucocorticoid therapies.

We used mouse ENCODE data along with complementary data from other laboratories to study the dynamics of occupancy and the role in gene regulation of the transcription factor TAL1, a critical regulator of hematopoiesis, at multiple stages of hematopoietic differentiation. We combined ChIP-seq and RNA-seq data in six mouse cell types representing a progression from multilineage precursors to differentiated erythroblasts and megakaryocytes. We found that sites of occupancy shift dramatically during commitment to the erythroid lineage, vary further during terminal maturation, and are strongly associated with changes in gene expression. In multilineage progenitors, the likely target genes are enriched for hematopoietic growth and functions associated with the mature cells of specific daughter lineages (such as megakaryocytes). In contrast, target genes in erythroblasts are specifically enriched for red cell functions. Furthermore, shifts in TAL1 occupancy during erythroid differentiation are associated with gene repression (dissociation) and induction (co-occupancy with GATA1). Based on both enrichment for transcription factor binding site motifs and co-occupancy determined by ChIP-seq, recruitment by GATA transcription factors appears to be a stronger determinant of TAL1 binding to chromatin than the canonical E-box binding site motif. Studies of additional proteins lead to the model that TAL1 regulates expression after being directed to a distinct subset of genomic binding sites in each cell type via its association with different complexes containing master regulators such as GATA2, ERG, and RUNX1 in multilineage cells and the lineage-specific master regulator GATA1 in erythroblasts.

We study the fundamental problem of pattern matching in the case where the string data is weighted: for every position of the string and every letter of the alphabet a probability of occurrence for this letter at this position is given. Sequences of this type are commonly used to represent uncertain data. They are of particular interest in computational molecular biology as they can represent different kind of ambiguities in DNA sequences: distributions of SNPs in genomes populations; position frequency matrices of DNA binding profiles; or even sequencing-related uncertainties. A weighted string may thus represent many different strings, each with probability of occurrence equal to the product of probabilities of its letters at subsequent positions. In this article, we present new average-case results on pattern matching on weighted strings and show how they are applied effectively in several biological contexts. A free open-source implementation of our algorithms is made available.

We study pattern matching problems on two major representations of uncertain sequences used in molecular biology: weighted sequences (also known as position weight matrices, PWM) and profiles (scoring matrices). In the simple version, in which only the pattern or only the text is uncertain, we obtain efficient algorithms with theoretically-provable running times using a variation of the lookahead scoring technique. We also consider a general variant of the pattern matching problems in which both the pattern and the text are uncertain. Central to our solution is a special case where the sequences have equal length, called the consensus problem. We propose algorithms for the consensus problem parameterised by the number of strings that match one of the sequences. As our basic approach, a careful adaptation of the classic meet-in-the-middle algorithm for the knapsack problem is used. On the lower bound side, we prove that our dependence on the parameter is optimal up to lower-order terms conditioned on the optimality of the original algorithm for the knapsack problem. Therefore, we make an effort to keep the lower order terms of the complexities of our algorithms as small as possible.

We study pattern matching problems on two major representations of uncertain sequences used in molecular biology: weighted sequences (also known as position weight matrices, PWM) and profiles (i.e., scoring matrices). In the simple version, in which only the pattern or only the text is uncertain, we obtain efficient algorithms with theoretically-provable running times using a variation of the lookahead scoring technique. We also consider a general variant of the pattern matching problems in which both the pattern and the text are uncertain. Central to our solution is a special case where the sequences have equal length, called the consensus problem. We propose algorithms for the consensus problem parameterized by the number of strings that match one of the sequences. As our basic approach, a careful adaptation of the classic meet-in-the-middle algorithm for the knapsack problem is used. On the lower bound side, we prove that our dependence on the parameter is optimal up to lower-order terms conditioned on the optimality of the original algorithm for the knapsack problem.

We study pattern matching on two major representations of uncertain sequences in computational biology---weighted sequences (also known as position weight matrices, PWM) and profiles (i.e., scoring matrices)---in connection with a multichoice generalization of the classic knapsack problem. We propose algorithms for profile matching and multichoice knapsack parameterized by the number of solutions with feasible weight. In the worst case, our algorithms run as fast as the best known non-parameterized solutions for these problems. For multichoice knapsack, this proves that our dependence on the parameter is optimal up to lower-order terms conditioned on the optimality of the state-of-the-art algorithms for Subset Sum and 3-Sum problems. We apply our algorithms to derive efficient solutions for the classic and general version of weighted pattern matching---in the former only the text is weighted, whereas in the latter both the pattern and the text can be weighted. We also argue that the resulting algorithm for general weighted pattern matching inherits the optimality results from multichoice knapsack.

We revisit the k-mismatch problem, one of the most basic problems in pattern matching. Given a pattern of length m and a text of length n, the task is to find all m-length substrings of the text that are at the Hamming distance at most $k$ from the pattern. Our first contribution is two new streaming algorithms for the k-mismatch problem. 1) For $k = 1$, we give an $O(polylog{m})$-space algorithm that uses $O(polylog{m})$ worst-case time per arriving symbol; 2) For $k \ge 2$, the problem can be solved in $O(k^2 polylog{m})$ space and $O(k \, polylog{m})$ worst-case time per arriving symbol. The complexities of our new k-mismatch algorithms match (up to logarithmic factors) those of the best known solutions from FOCS 2009 and SODA 2016, and our algorithms are enhanced with an important new feature which we call Data Recovery: For each alignment where the Hamming distance is at most k, the algorithms output the difference of symbols of the pattern and of the text in the mismatching positions. This is a particularly surprising feature as we are not allowed to store a copy of the pattern or the text in the streaming setting. Armed with this new feature, we develop the first streaming algorithms for pattern matching on weighted strings. Unlike regular strings, a weighted string is not a sequence of symbols but rather a sequence of probability distributions on the alphabet. In the Weighted Pattern Matching problem we are given a text and a pattern, both of which can be either weighted or regular strings, and must find all alignments of the text and of the pattern where they match with probability above a given threshold. The algorithms presented in the paper are randomised and correct with high probability.

We report the genome-wide identification of estrogen receptor α (ERα)-binding regions in mouse liver using a combination of chromatin immunoprecipitation and tiled microarrays that cover all nonrepetitive sequences in the mouse genome. This analysis identified 5568 ERα-binding regions. In agreement with what has previously been reported for human cell lines, many ERα-binding regions are located far away from transcription start sites; approximately 40% of ERα-binding regions are located within 10 kb of annotated transcription start sites. Almost 50% of ERα-binding regions overlap genes. The majority of ERα-binding regions lie in regions that are evolutionarily conserved between human and mouse. Motif-finding algorithms identified the estrogen response element, and variants thereof, together with binding sites for activator protein 1, basic-helix-loop-helix proteins, ETS proteins, and Forkhead proteins as the most common motifs present in identified ERα-binding regions. To correlate ERα binding to the prom...

We report the generation and analysis of functional data from multiple, diverse experiments performed on a targeted 1% of the human genome as part of the pilot phase of the ENCODE Project. These data have been further integrated and augmented by a number of evolutionary and computational analyses. Together, our results advance the collective knowledge about human genome function in several major areas. First, our studies provide convincing evidence that the genome is pervasively transcribed, such that the majority of its bases can be found in primary transcripts, including non-protein-coding transcripts, and those that extensively overlap one another. Second, systematic examination of transcriptional regulation has yielded new understanding about transcription start sites, including their relationship to specific regulatory sequences and features of chromatin accessibility and histone modification. Third, a more sophisticated view of chromatin structure has emerged, including its inter-relationship with DNA replication and transcriptional regulation. Finally, integration of these new sources of information, in particular with respect to mammalian evolution based on inter- and intra-species sequence comparisons, has yielded new mechanistic and evolutionary insights concerning the functional landscape of the human genome. Together, these studies are defining a path for pursuit of a more comprehensive characterization of human genome function.

We report the association of an inherited variant located upstream of the poly(adenosine diphosphate‐ribose) polymerase 1 (PARP1) gene (rs2249844), with survival in 11 BioGenoMEL melanoma cohorts. The gene encodes a protein involved in a number of cellular processes including single‐strand DNA repair. Survival analysis was conducted for each cohort using proportional hazards regression adjusting for factors known to be associated with survival. Survival was measured as overall survival (OS) and, where available, melanoma‐specific survival (MSS). Results were combined using random effects meta‐analysis. Evidence for a role of the PARP1 protein in melanoma ulceration and survival was investigated by testing gene expression levels taken from formalin‐fixed paraffin‐embedded tumors. A significant association was seen for inheritance of the rarer variant of PARP1, rs2249844 with OS (hazard ratio (HR) = 1.16 per allele, 95% confidence interval (CI) 1.04–1.28, p = 0.005, eleven cohorts) and MSS (HR = 1.20 per allele, 95% CI 1.01–1.39, p = 0.03, eight cohorts). We report bioinformatic data supportive of a functional effect for rs2249844. Higher levels of PARP1 gene expression in tumors were shown to be associated with tumor ulceration and poorer OS.

We present the second and improved release of the TOUCAN workbench for cis-regulatory sequence analysis. TOUCAN implements and integrates fast state-of-the-art methods and strategies in gene regulation bioinformatics, including algorithms for comparative genomics and for the detection of cis-regulatory modules. This second release of TOUCAN has become open source and thereby carries the potential to evolve rapidly. The main goal of TOUCAN is to allow a user to come to testable hypotheses regarding the regulation of a gene or of a set of co-regulated genes. TOUCAN can be launched from this location: .

We present probabilistic arithmetic automata (PAAs), which can be used to model chains of operations whose operands depend on chance. We provide two different algorithms to exactly calculate the distribution of the results obtained by such probabilistic calculations. Although we introduce PAAs and the corresponding algorithm in a generic manner, our main concern is their application to pattern matching statistics, i.e. we study the distributions of the number of occurrences of a pattern under a given text model. Such calculations play an important role in computational biology as they give access to the significance of pattern occurrences. To assess the practicability of our method, we apply it to the Prosite database of amino acid motifs and to the Jaspar database of transcription factor binding sites. Regarding the latter, we additionally show that our framework permits to take binding affinities predicted from a physical model into account.

We present ensemble methods in a machine learning (ML) framework combining predictions from five known motif/binding site exploration algorithms. For a given TF the ensemble starts with position weight matrices (PWM's) for the motif, collected from the component algorithms. Using dimension reduction, we identify significant PWM-based subspaces for analysis. Within each subspace a machine classifier is built for identifying the TF's gene (promoter) targets (Problem 1). These PWM-based subspaces form an ML-based sequence analysis tool. Problem 2 (finding binding motifs) is solved by agglomerating k-mer (string) feature PWM-based subspaces that stand out in identifying gene targets. We approach Problem 3 (binding sites) with a novel machine learning approach that uses promoter string features and ML importance scores in a classification algorithm locating binding sites across the genome. For target gene identification this method improves performance (measured by the F1 score) by about 10 percentage points over the (a) motif scanning method and (b) the coexpression-based association method. Top motif outperformed 5 component algorithms as well as two other common algorithms (BEST and DEME). For identifying individual binding sites on a benchmark cross species database (Tompa et al., 2005) we match the best performer without much human intervention. It also improved the performance on mammalian TFs. The ensemble can integrate orthogonal information from different weak learners (potentially using entirely different types of features) into a machine learner that can perform consistently better for more TFs. The TF gene target identification component (problem 1 above) is useful in constructing a transcriptional regulatory network from known TF-target associations. The ensemble is easily extendable to include more tools as well as future PWM-based information.

We present an intuitive and interactive web service for CARRIE (Computational Ascertainment of Regulatory Relationships Inferred from Expression). CARRIE is a computational method that analyzes microarray and promoter sequence data to infer a transcriptional regulatory network from the response to a specific stimulus. This service displays an interactive graph of the inferred network and provides easy access to the evidence for the involvement of each gene in the network. We provide functionality to include network data in KEGG XML (KGML) format in this graph. Our service also provides Gene Ontology annotation to aid the user in forming hypotheses about the role of each gene in the cellular response. The CARRIE web service is freely available at http://zlab.bu.edu/CARRIE-web.

We present a new streaming algorithm for the k-Mismatch problem, one of the most basic problems in pattern matching. Given a pattern and a text, the task is to find all substrings of the text that are at the Hamming distance at most k from the pattern. Our algorithm is enhanced with an important new feature called Error Correcting, and its complexities for k = 1 and for a general k match those of the best known solutions for the k-Mismatch problem from FOCS 2009 and SODA 2016. As a corollary we develop a series of streaming algorithms for pattern matching on weighted strings, which are a commonly used representation of uncertain sequences in molecular biology.

We present a computational method aimed at systematically identifying tissue-selective transcription factor binding sites. Our method focuses on the differences between sets of promoters that are associated with differentially expressed genes, and it is effective at identifying the highly degenerate motifs that characterize vertebrate transcription factor binding sites. Results on simulated data indicate that our method detects motifs with greater accuracy than the leading methods, and its detection of strongly overrepresented motifs is nearly perfect. We present motifs identified by our method as the most overrepresented in promoters of liver- and muscle-selective genes, demonstrating that our method accurately identifies known transcription factor binding sites and previously uncharacterized motifs.