RNA molecules play important roles, including catalysis of chemical reactions and control of gene expression, and their functions largely depend on their folded structures. Since determining these structures by biochemical means is expensive, there is increased demand for computational predictions of RNA structures. One computational approach is to find the secondary structure (a set of base pairs) that minimizes a free energy function for a given RNA conformation. The forces driving RNA folding can be approximated by means of a free energy model, which associates a free energy parameter to a distinct considered feature. The main goal of this thesis is to develop state-of-the-art computational approaches that can significantly increase the accuracy (i.e., maximize the number of correctly predicted base pairs) of RNA secondary structure prediction methods, by improving and refining the parameters of the underlying RNA free energy model. We propose two general approaches to estimate RNA free energy parameters. The Constraint Generation (CG) approach is based on iteratively generating constraints that enforce known structures to have energies lower than other structures for the same molecule. The Boltzmann Likelihood (BL) approach infers a set of RNA free energy parameters which maximize the conditional likelihood of a set of known RNA structures. We discuss several variants and extensions of these two approaches, including a linear Gaussian Bayesian network that defines relationships between features. Overall, BL gives slightly better results than CG, but it is over ten times more expensive to run. In addition, CG requires software that is much simpler to implement. We obtain significant improvements in the accuracy of RNA minimum free energy secondary structure prediction with and without pseudoknots (regions of non-nested base pairs), when measured on large sets of RNA molecules with known structures. For the Turner model, which has been the gold-standard model without pseudoknots for more than a decade, the average prediction accuracy of our new parameters increases from 60% to 71%. For two models with pseudoknots, we obtain an increase of 9% and 6%, respectively. To the best of our knowledge, our parameters are currently state-of-the-art for the three considered models. / Science, Faculty of / Computer Science, Department of / Graduate

Sus1 is a conserved protein involved in chromatin remodeling and mRNA biogenesis. The SUS1 gene of Saccharomyces cerevisiae is unusual, as it contains two introns and is alternatively spliced, retaining one or both introns in response to changes in environmental conditions. SUS1 splicing may allow the cell to control Sus1 expression, but the mechanisms that regulate this process remain unknown. In this thesis project, we have investigated whether the structure adopted by SUS1 RNA sequences contributes to regulate the splicing of this gene. Using in silico analyses together with NMR spectroscopy, gel electrophoresis and UV thermal denaturation experiments, we first show that the downstream intron (I2) of SUS1 forms a weakly-stable, 37-nucleotide stem-loop structure containing the branch site near its apical loop and the 3' splice site after the stem terminus. A cellular assay revealed that two of four mutants containing altered I2 structures had significantly impaired SUS1 expression. Semi-quantitative RT-PCR experiments indicated that all mutants accumulated unspliced SUS1 pre-mRNA and/or induced distorted levels of fully spliced mRNA relative to wild-type. Concomitantly, Sus1 cellular functions in histone H2B deubiquitination and mRNA export were affected in I2 hairpin mutants that inhibited splicing. The second part of the thesis project focuses on the exon located between the two introns of the SUS1 gene. This middle exon (E2) can be skipped during splicing, is generated in circular form, and has been found to influence the splicing of the flanking introns, an unusual situation in budding yeast where splicing mainly relies on intron recognition. Using NMR spectroscopy, gel electrophoresis, UV thermal denaturation and ribose 2'-OH modification experiments combined with computational predictions, we show that E2 of SUS1 comprises a conserved double-helical stem topped by a three-way junction. One of the hairpins emerging from the junction exhibited significant thermal stability and was closed by an unusually structured purine-rich loop. This loop contained two consecutive sheared G:A base pairs and was structurally related to the substrate loop of the VS ribozyme. Cellular assays revealed that three mutants containing altered E2 structures had impaired SUS1 expression and that a compensatory mutation restoring the conserved stem recovered expression to wild-type levels. Semi-quantitative RT-PCR experiments indicated that all mutants were capable of altering the quantities of unspliced and/or fully-spliced SUS1 RNA transcripts relative to wildtype. Overall, the results gathered in this thesis project indicate that RNA structures formed by the middle exon and the second intron of the S. cerevisiae SUS1 gene are relevant for splicing and also influence other processes of SUS1 mRNA biogenesis. / Sus1 es una proteína conservada implicada en remodelación de cromatina y biogénesis de moléculas de ARNm. El gen SUS1 de Saccharomyces cerevisiae es peculiar, ya que contiene dos intrones y sufre un proceso de ayuste (corte y empalme) alternativo, reteniendo uno o ambos intrones en respuesta a cambios en las condiciones ambientales. El ayuste del ARNpre-m de SUS1 puede permitir a la célula controlar la expresión de la proteína Sus1, pero los mecanismos que regulan este proceso son poco conocidos. En este proyecto de tesis hemos investigado si la estructura adoptada por secuencias de ARN de SUS1 contribuye a regular el proceso de ayuste de este gen. Utilizando análisis in silico junto con espectroscopia de RMN, electroforesis en gel y experimentos de desnaturalización térmica monitorizados por UV, primero demostramos que el ARN del segundo intrón (I2) del gen SUS1 forma una horquilla débilmente estable de 37 nucleótidos. Esta horquilla contiene nucleótidos del sitio de ramificación (branch site) en su bucle apical y nucleótidos del sitio 3' de empalme adyacentes al extremo inferior del tallo. A través de ensayos funcionales descubrimos que dos de cuatro mutantes que alteran la estructura de la horquilla I2 exhibían peor expresión de SUS1. Experimentos de RT-PCR semicuantitativos indicaron que todos los mutantes acumularon ARNpre-m SUS1 no ayustado y/o indujeron cambios en los niveles de ARNm maduro con respecto a la secuencia silvestre. Además, las funciones celulares de Sus1 relativas a desubicuitinación de histona H2B y transporte de ARNm se vieron afectadas en los mutantes de la horquilla I2 que inhibían el proceso de ayuste. La segunda parte de la memoria de tesis se centra en el análisis del exón central (E2) situado entre los dos intrones del gen SUS1. Este exón puede eliminarse durante el proceso de ayuste, se genera en forma circular, e influye en el procesamiento de los intrones adyacentes, una situación inusual para las regiones exónicas de S. cerevisiae, donde el ayuste se basa principalmente en el reconocimiento de intrones. Utilizando experimentos de espectroscopía de RMN, electroforesis en gel, desnaturalización térmica y modificación química combinados con predicciones computacionales, demostramos que el ARN del exón E2 de SUS1 forma un tallo conservado de doble hélice coronado por una intersección de tres hélices. Una de las horquillas que emergen de esta intersección presentó una estabilidad térmica significativa, así como un bucle apical rico en purinas inusualmente estructurado. Este bucle contiene dos pares de bases G:A consecutivos y está estructuralmente relacionado con el bucle de substrato de la ribozima VS. Ensayos celulares revelaron que tres mutantes con estructuras modificadas de E2 exhibían peor expresión de SUS1, y que una mutación compensatoria que restauraba el tallo conservado recuperaba la expresión a los niveles de la secuencia silvestre. Experimentos de RT-PCR semicuantitativos indicaron que todos los mutantes de E2 eran capaces de alterar las cantidades de transcritos ayustados y no ayustados de SUS1 con respecto a la secuencia silvestre. En general, los resultados obtenidos en este proyecto de tesis indican que las estructuras de ARN formadas por el exón central y el segundo intrón del gen SUS1 de S. cerevisiae son relevantes para el ayuste y otros procesos implicados en la biogénesis del ARNm del gen SUS1. / Sus1 és una proteïna conservada implicada a la remodelació de la cromatina i la biogènesi de l'ARNm. El gen SUS1 de Saccharomyces cerevisiae és inusual, ja que conté dos introns i s'empalma de manera alternativa, retenint un o ambdós introns en resposta a canvis en les condicions ambientals. L'empalmament de SUS1 pot permetre a la cèl·lula controlar l'expressió de Sus1, però els mecanismes que regulen aquest procés són segueixen sent desconeguts. En aquest projecte de tesi investiguem si l'estructura adoptada per seqüències d'ARN de SUS1 contribueix a regular l'empalmament d'aquest gen. Emprant anàlisi in silico juntament amb espectrometria de RMN, electroforesi en gel i experiments de desnaturalització tèrmica d'UV, es mostra primer que l'intró aigües a baix (I2) de SUS1 forma una estructura de forqueta de 37 nucleòtids feblement estable que conté el lloc de la branca a prop del seu bucle apical; i el lloc d'empalmamnet 3¿ després de l'extrem de la forqueta. Un assaig cel·lular va revelar que dos de quatre mutants que contenien estructures alterades de l'I2 havien modificat significativament l'expressió de SUS1. Els experiments semi-quantitatius de RT-PCR van indicar que tots els mutants acumulaven el pre-ARNm madur respecte al tipus salvatge. Concomitantment, les funcions cel·lulars de Sus1 a la desubiqüitinació de la histona H2B i l'exportació d'ARNm es van veure afectats als mutants de la forqueta d'I2 que inhibeixen l'empalmament. La segona part del projecte de tesi se centra a l'exó situat entre els dos introns del gen SUS1. Aquest exó (E2) es pot ometre durant l'empalmament, es genera amb forma circular, i s'ha trobat que influeix a l'empalmamet dels introns que flanquegen, una situació inusual al llevat on l'empalmament està basat principalment al reconeixement d'introns. Emprant espectroscòpia de RMN, electroforesi en gel, desnaturalització tèrmica d'UV i experiments de modificació de ribosa 2¿-OH combinats amb prediccions computacionals, mostrem que E2 de SUS1 comprén un tall conservat de doble hèlix corornat per una unió de tres vies. Una de les forquetes que emergeixen de la unió, va mostrar una estabilitat tèrmica significativa i va ser tapada per un bucle ric en purina inusualment estructurat. Aquest bucle contenia dos pars de bases G:A tallats consecutivament i estava estructuralment relacionat amb el bucle de substrat del ribozim VS. Els assajos cel·lulars van revelar que tres mutants que contenien estructures alterades de E2 havien alterat l'expressió de SUS1 i que una mutació compensatòria que restaurava el tall conservat recuperava l'expressió a nivells del tipus salvatge. Els assajos cel·lulars van revelat que tres mutants que contenien estructures alterades d'E2 havien alterat l'expressió de SUS1 i que una mutació compensatòria que restaurava el tall conservat recuperava l'expressió a nivell d'un tipus salvatge. Els experiments semi-quantitatius de RT-PCR van indicar que tots els mutants eren capaços d'alterar les quantitats de transcrits d'ARN de SUS1 no empalmats i/o empalmats en relació amb el tipus salvatge. En general, els resultats obtinguts en aquesta investigació indiquen que les estructures d'ARN formades per l'exó mitjà i el segon intró de SUS1 de S. cerevisiae són rellevants per l'empalmament i també influeixen a altres processos de biogènesi de l'ARN de SUS1 / Abuqattam, AN. (2017). IDENTIFICATION OF RNA STRUCTURES MODULATING THE EXPRESSION OF THE mRNA BIOGENESIS FACTOR SUS1 [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/90524 / TESIS

We extend earlier cellular automata models of spatially extended hypercycles by including an explicit genetic component into the model. This allows us to study the sequence evolution of hypercyclically coupled molecular replicators in addition to considering their population dynamics and spatial organization. In line with previous models, that considered either spatial organization or sequence evolution alone, we find both temporal oscillations of the relative concentration of the species forming the hypercycles as well as the formation of spatial organisations including spiral waves. We also confirm the greatly increased robustness of the spatially extended hypercycle against various classes of parasites. We find the sequence evolution of each of the hypercyclically coupled populations proceeds (after an inital selection-dominated phase) in a drift-like manner that can be described by a diffusion process in sequence space. Kimura's theory of neutral evolution is therefore applicable on long time-scales despite the fact that the hypercycle exhibits extreme periodic changes in population sizes and that are governed solely by frequency-dependent selection.

RNAs are involved in different facets of biological processes; including but not limited to controlling and inhibiting gene expressions, enabling transcription and translation from DNA to proteins, in processes involving diseases such as cancer, and virus-host interactions. As such, there are useful applications that may arise from studies and analyses involving RNAs, such as detecting cancer by measuring the abundance of specific RNAs, detecting and identifying infections involving RNA viruses, identifying the origins of and relationships between RNA viruses, and identifying potential targets when designing novel drugs. Extracting sequences from RNA samples is usually not a major limitation anymore thanks to sequencing technologies such as RNA-Seq. However, accurately identifying and analyzing the extracted sequences is often still the bottleneck when it comes to developing RNA-based applications. Like proteins, functional RNAs are able to fold into complex structures in order to perform specific functions throughout their lifecycle. This suggests that structural information can be used to identify or classify RNA sequences, in addition to the sequence information of the RNA itself. Furthermore, a strand of RNA may have more than one possible structural conformations it can fold into, and it is also possible for a strand to form different structures in vivo and in vitro. However, past studies that utilized secondary structure information for RNA identification purposes have relied on one predicted secondary structure for each RNA sequence, despite the possible one-to-many relationship between a strand of RNA and the possible secondary structures. Therefore, we hypothesized that using a representation that includes the multiple possible secondary structures of an RNA for classification purposes may improve the classification performance. We proposed and built a pipeline that produces secondary structure fingerprints given a sequence of RNA, that takes into account the aforementioned multiple possible secondary structures for a single RNA. Using this pipeline, we explored and developed different types of secondary structure fingerprints in our studies. A type of fingerprints serves as high-level topological representations of the RNA structure, while another type represents matches with common known RNA secondary structure motifs we have curated from databases and the literature. Next, to test our hypothesis, the different fingerprints are then used with deep learning and with different datasets, alone and together with various sequence-based features, to investigate how the secondary structure fingerprints affect the classification performance. Finally, by analyzing our findings, we also propose approaches that can be adopted by future studies to further improve our secondary structure fingerprints and classification performance.

This thesis develops probabilistic models of RNA secondary structure. The first chapter introduces RNA secondary structure prediction, in particular stochastic context-free grammars (SCFGs), and considers a novel method for automated design of SCFGs. Many SCFGs are found with a similar predictive quality as those commonly used for RNA secondary structure prediction. The second chapter discusses the effect alignment quality, evolutionary distance between sequences, and number of sequences in an alignment have on RNA secondary structure prediction. By combining statistical alignment and SCFG models we can, in a statistically sound setting, average structure predictions over the space of alignments to decrease loss created by poor alignments. The third chapter incorporates additional biological information about RNA secondary structure formation into the decoding of the SCFG posterior distribution. Combining iterative helix formation, phylogenetic modelling, and a distance function between alignment columns leads to the an improvement in the accuracy of comparative RNA secondary structure prediction. Finally, appendices briefly discuss further work concerning probabilistic models of RNA secondary structure which may be of interest to the reader.

In this dissertation, we explore the transcriptional regulatory roles of Gquadruplex- forming motifs and the involvement of specific transcriptional factors, which interact with the same elements, in the control of human c-myc and bcl-2 gene expression. The G-quadruplex structures within the NHE III1 region of the c-myc promoter and their ability to repress transcription has been well established. However, a longstanding unanswered question is how these stable DNA secondary structures are transformed to activate c-myc transcription. NDPK-B has been recognized as an activator of c-myc transcription via interactions with NHE III1 region of the c-myc gene promoter. Through the use of RNAi, we confirmed the transcriptional regulatory role of NDPK-B. We demonstrate that NDPK-B has DNA binding activity and the nuclease activity results from a contaminating protein. NDPK-B preferentially binds to the singlestranded guanine-rich strand of the c-myc NHE III₁. Potassium ions and G-quadruplexinteractive agents, which stabilize G-quadruplex structures, had an inhibitory effect on NDPK-B DNA binding activity. Based on our studies, we have proposed a stepwise trapping-out of the NHE III1 region in a single-stranded form, thus allowing singlestranded transcription factors to bind and activate c-myc transcription. This model provides a rationale for how the stabilization of G-quadruplexes within the c-myc gene promoter region can inhibit NDPK-B from activating c-myc transcription. Similarly, the human bcl-2 gene contains a GC-rich region within its promoter region, which is critical in the regulation of bcl-2 expression. We demonstrate that the guanine rich strand within this region can form three intramolecular G-quadruplex structures. Based on NMR studies, the central G-quadruplex forms a mixed parallel/antiparallel structure with three tetrads connected by loops of one, seven, and three bases. The Gquadruplex structures in the bcl-2 promoter extends beyond the ability to form any one of three separate G-quadruplexes to each having the capacity to form either three or six different loop isomers. This suggests that targeting these individual structures could lead to different biological outcomes. We also found that Telomestatin upregulates bcl-2 gene expression, which we propose is a result of inhibiting the binding of the WT1 repressor protein by the formation of a drug-stabilized G-quadruplex structure.

A phenomenon as complex as protein folding requires a complex model to approximate it. This thesis presents a bottom-up approach for building complex probabilistic models of protein secondary structure by incorporating the multiple information sources which we call experts. Expert opinions are represented by probability distributions over the set of possible structures. Bayesian treatment of a group of experts results in a consensus opinion that combines the experts' probability distributions using the operators of normalized product, quotient and exponentiation. The expression of this consensus opinion simplifiesto a product of the expert opinions with two assumptions: (1) balanced training of experts, i. e. , uniform prior probability over all structures, and (2) conditional independence between expert opinions,given the structure. This research also studies how Markov chains and hidden Markov models may be used to represent expert opinion. Closure properties areproven, and construction algorithms are given for product of hidden Markov models, and product, quotient and exponentiation of Markovchains. Algorithms for extracting single-structure predictions from these models are also given. Current product-of-experts approaches in machine learning are top-down modeling strategies that assume expert independence, and require simultaneous training of all experts. This research describes a bottom-up modeling strategy that can incorporate conditionally dependent experts, and assumes separately trained experts.

Thesis advisor: Peter Clote / In addition to the well-characterized messenger RNA, transfer RNA and ribosomal RNA, many new classes of noncoding RNA(ncRNA) have been discovered in the past few years. ncRNA has been shown to play important roles in multiple regulation and development processes. The increasing needs for RNA structural analysis software provide great opportunities on computational biologists. In this thesis I present three highly non-trivial RNA parametric structural analysis algorithms: 1) RNAhairpin and RNAmultiloop, which calculate parition functions with respect to hairpin number, multiloop number and multiloop order, 2) RNAshapeEval, which is based upon partition function calculation with respect to a fixed abstract shape, and 3) RNAprofileZ, which calculates the expected partition function and ensemble free energy given an RNA position weight matrix.I also describe the application of these software in biological problems, including evaluating purine riboswitch aptamer full alignment sequences to adopt their consensus shape, building hairpin and multiloop profiles for certain Rfam families, tRNA and pseudoknotted RNA secondary structure predictions. These algorithms hold the promise to be useful in a broad range of biological problems such as structural motifs search, ncRNA gene finders, canonical and pseudoknotted secondary structure predictions. / Thesis (MS) — Boston College, 2010. / Submitted to: Boston College. Graduate School of Arts and Sciences. / Discipline: Biology.

MicroRNAs are important biomolecules for regulating biological processes. Moreover, the secondary structure of microRNA is important for its activity and has been used previously as a mean for finding unknown microRNAs. A phylogenetic study of the microRNA secondary structure reveals more information than its primary sequence, because the primary sequence can undergo mutations that give rise to different phylogenetic relationships, whereas the secondary structure is more robust against mutations and therefore sometimes  more informative. Here we constructed a phylogenetic tree entirely based on microRNA secondary structures using tools PHYLIP (Felsenstein, 1995) and RNAforester (Matthias Höchsmann, 2003, Hochsmann et al., 2004), and compared the overall topology and clusters with the phylogenetic tree constructed using microRNA sequence. The purpose behind this comparison was to investigate the sequence and structure similarity in phylogenetic context and also to investigate if functionally similar microRNA genes are closer in their structure-derived phylogenetic tree. Our phylogenetic comparison shows that the sequence similarity has hardly any effect on the structure similarity in the phylogenetic tree. MicroRNAs that have similar function are closer in the phylogenetic tree based on secondary structure than its respective sequence phylogeny. Hence, this approach can be very useful in predicting the functions of the new microRNAs whose function is yet to be known, since the function of the miRNAs heavily relies on its secondary structure.

Understanding structure and folding of a protein is the key to understanding its biological function and potential role in diseases. Despite the importance of protein folding, a molecular level understanding of this process is still lacking. Solution-state nuclear magnetic resonance (NMR) is a powerful technique to investigate protein structure, dynamics, and folding mechanisms, since it provides residue-specific information. One of the major contributions that govern protein structure appears to be the interaction with the solvent. The importance of these interactions is particularly apparent in membrane proteins, which exist in an amphiphilic environment. Here, individual peptide fragments taken from the disulfide bond forming protein B (DsbB) were investigated in various solvents. The alpha-helical structures that were obtained, suggest that DsbB follows the two-stage model for folding. However, side chains of polar residues showed different conformations compared to the X-ray structure of fulllength protein, implying that polar side-chains may re-orient upon helix packing in order to form the necessary tertiary interactions that stabilize the global fold of DsbB. Model peptides in general represent attractive systems for the investigation of non-covalent interactions important for protein folding, including those with the solvent. NMR structures of the water soluble peptide, BBA5, were obtained in the presence an organic co-solvent, methanol. These structures indicate that the addition of methanol stabilizes an alpha-helix segment, but disrupts a hydrophobic cluster forming a beta-hairpin. Since dynamic effects reduce the ability for experimental observation of individual, bound solvent molecules, results were compared with molecular dynamics simulations. This comparison indicates that the observed effects of NMR structures are due to preferred binding of methanol and reduction of peptide-water hydrogen bonding. NMR structures, such as those determined here, represent a distribution of conformations under equilibrium. The dynamic process of protein unfolding can nevertheless be accessed through denaturation. A method was developed to probe thermal denaturation by measuring the temperature dependence of NOE intensity. Applied to a model peptide, trpzip4, it was confirmed that the beta-hairpin structure of this peptide is stabilized by the hydrophobic cluster formed by tryptophan residues. Together, the peptides investigated here illustrate the important roles that solvent-peptide interactions and side chain-side chain hydrophobic interactions play in forming stable secondary and tertiary structures.