A number of recent studies have shown that the dimensionality of the neural activity in the cortex is low. However, what network structures are capable of producing such activity is not theoretically well understood. In this thesis, I discuss a few possible solutions to this problem, and demonstrate that a network with a multidimensional attractor can give rise to such low dimensional activity. The network is created using the Neural Engineering Framework, and exhibits several biologically plausible features, including a log-normal distribution of the synaptic weights. / Ett antal nyligen publicerade studier has visat att dimensionaliten för neural aktivitet är låg. Dock är det inte klarlagt vilka nätverksstrukturer som kan uppbringa denna typ av aktivitet. I denna uppsats diskuterar jag möjliga lösningsförslag, och demonstrerar att ett nätverk med en flerdimensionell attraktor ger upphov till lågdimensionell aktivitet. Nätverket skapas med hjälp av the Neural Engineering Framework, och uppvisar ett flertal biologiskt trovärdiga egenskaper. I synnerhet är fördelningen av synapsvikter log-normalt fördelad.

Over the last decade, development and application of a set of molecular genomic approaches based on the chromosome conformation capture method (3C), combined with increasingly powerful imaging approaches have enabled high resolution and genome-wide analysis of the spatial organization of chromosomes. The aim of this thesis is two-fold; 1), to provide guidelines for analyzing and interpreting data obtained from genome-wide 3C methods such as Hi-C and 3C-seq and 2), to leverage the 3C technology to solve genome function, structure, assembly, development and dosage problems across a broad range of organisms and disease models. First, through the introduction of cWorld, a toolkit for manipulating genome structure data, I accelerate the pace at which *C experiments can be performed, analyzed and biological insights inferred. Next I discuss a set of practical guidelines one should consider while planning an experiment to study the structure of the genome, a simple workflow for data processing unique to *C data and a set of considerations one should be aware of while attempting to gain insights from the data. Next, I apply these guidelines and leverage the cWorld toolkit in the context of two dosage compensation systems. The first is a worm condensin mutant which shows a reduction in dosage compensation in the hermaphrodite X chromosomes. The second is an allele-specific study consisting of genome wide Hi-C, RNA-Seq and ATAC-Seq which can measure the state of the active (Xa) and inactive (Xi) X chromosome. Finally I turn to studying specific gene – enhancer looping interactions across a panel of ENCODE cell-lines. These studies, when taken together, further our understanding of how genome structure relates to genome function.

Analysis of gene expression has undergone a technological revolution. What was impossible 6 years ago is now routine. High-throughput DNA sequencing machines capable of generating hundreds of millions of reads allow, indeed force, a major revision toward the study of the genome’s functional output—the transcriptome. This thesis examines the history of DNA sequencing, measurement of gene expression by sequencing, isoform complexity driven by alternative splicing and mammalian piRNA precursor biogenesis. Examination of these topics is framed around development of a novel RNA-templated DNA-DNA ligation assay (SeqZip) that allows for efficient analysis of abundant, complex, and functional long RNAs. The discussion focuses on the future of transcriptome analysis, development and applications of SeqZip, and challenges presented to biomedical researchers by extremely large and rich datasets.

The human immunodeficiency virus type 1 (HIV-1) protease (PR) is a critical drug target as it is responsible for virion maturation. Mutations within the active site (1°) of the PR directly interfere with inhibitor binding while mutations distal to the active site (2°) to restore enzymatic fitness. Increasing mutation number is not directly proportional to the severity of resistance, suggesting that resistance is not simply additive but that it is interdependent. The interdependency of both primary and secondary mutations to drive protease inhibitor (PI) resistance is grossly understudied. To structurally and dynamically characterize the direct role of secondary mutations in drug resistance, I selected a panel of single-site mutant protease crystal structures complexed with the PI darunavir (DRV). From these studies, I developed a network hypothesis that explains how mutations outside the active site are able to perpetuate changes to the active site of the protease to disrupt inhibitor binding. I then expanded the panel to include highly mutated multi-drug resistant variants. To elucidate the interdependency between primary and secondary mutations I used statistical and machine-learning techniques to determine which specific mutations underlie the perturbations of key inter-molecular interactions. From these studies, I have determined that mutations distal to the active site are able to perturb the global PR hydrogen bonding patterns, while primary and secondary mutations cooperatively perturb hydrophobic contacts between the PR and DRV. Discerning and exploiting the mechanisms that underlie drug resistance in viral targets could proactively ameliorate both current treatment and inhibitor design for HIV-1 targets.

Small RNAs are single-stranded, 18–36 nucleotide RNAs that can be categorized as miRNA, siRNA, and piRNA. miRNA are expressed ubiquitously in tissues and at particular developmental stages. They fine-tune gene expression by regulating the stability and translation of mRNAs. piRNAs are mainly expressed in the animal gonads and their major function is repressing transposable elements to ensure the faithful transfer of genetic information from generation to generation. My thesis research focused on the biogenesis of miRNAs and piRNAs using both experimental and computational strategies. The biogenesis of miRNAs involves sequential processing of their precursors by the RNase III enzymes Drosha and Dicer to generate miRNA/miRNA* duplexes, which are subsequently loaded into Argonaute proteins to form the RNA-induced silencing complex (RISC). We discovered that, after assembled into Ago1, more than a quarter of Drosophila miRNAs undergo 3′ end trimming by the 3′-to-5′ exoribonuclease Nibbler. Such trimming occurs after removal of the miRNA* strand from pre-RISC and may be the final step in RISC assembly, ultimately enhancing target messenger RNA repression. Moreover, by developing a specialized Burrow-Wheeler Transform based short reads aligner, we discovered that in the absence of Nibbler a subgroup of miRNAs undergoes increased tailing—non-templated nucleotide addition to their 3′ ends, which are usually associated with miRNA degradation. Therefore, the 3′ trimming by Nibbler might increase miRNA stability by protecting them from degradation. In Drosophila germ line, piRNAs associate with three PIWI-clade Argonaute proteins, Piwi, Aub, and Ago3. piRNAs bound by Aub and Ago3 are generated by reciprocal cleavages of sense and antisense transposon transcripts (a.k.a., the “Ping-Pong” cycle), which amplifies piRNA abundance and degrades transposon transcripts in the cytoplasm. On the other hand, Piwi and its associated piRNA repress the transcription of transposons in the nucleus. We discovered that Aub- and Ago3-mediated transposon RNA cleavage not only generates piRNAs bound to each other, but also produces substrates for the endonuclease Zucchini, which processively cleaves those substrates in a periodicity of ~26 nt and generates piRNAs that predominantly load into Piwi. Without Aub or Ago3, the abundance of Piwi-bound piRNAs drops and transcriptional silencing is compromised. Our discovery revises the current model of piRNA biogenesis.

Viruses are known to be associated with 20% of human cancers. Epstein Barr virus (EBV) in particular is the first virus associated with human cancers. Here, we computationally detect EBV and explore the effects of this virus across cancers by taking advantage of the fact that EBV microRNAs (miRNAs) and Epstein Barr virus small RNAs (EBERs) are expressed at all viral latencies. We identify and characterize two sub-populations of EBV positive tumors: those with high levels of EBV miRNA and EBERS expression and those with medium levels of expression. Based on principal component analysis (PCA) and hierarchical clustering of viral miRNAs across all samples we observe a pattern of expression for these EBV miRNAs which is correlated with both the tumor cell type (B cell versus epithelial cell) and with the overall levels of expression of these miRNAs. We further investigated the effect of the levels of EBV miRNAs with the overall survival of patients across cancers. Through Kaplan Meier survival analysis we observe a significant correlation with levels of EBV miRNAs and lower survival in adult AML patients. We also designed a machine learning model for risk assessment of EBV in association with adult AML and other clinical factors. Our next aim was to identify targets of EBV miRNAs, hence, we used a combination of previously known methodologies for miRNA target detection in addition to a multivariable regression approach to identify targets of these viral miRNAs in stomach cancer. Finally, we investigate the variations across EBV subtype specific EBNA3C gene which interacts with the host immune system. Preliminary data suggests potential regional variations plus higher pathogenicity of subtype 1 in comparison to subtype 2 EBV. Overall, these studies further our understanding of how EBV manipulates the tumor microenvironment across cancer subtypes.

PIWI-interacting RNAs (piRNA) are a group of 23–35 nucleotide (nt) short RNAs that protect animal gonads from transposon activities. In Drosophila germ line, piRNAs can be categorized into two different categories— primary and secondary piRNAs— based on their origins. Primary piRNAs, generated from transcripts of specific genomic regions called piRNA clusters, which are enriched in transposon fragments that are unlikely to retain transposition activity. The transcription and maturation of primary piRNAs from those cluster transcripts are poorly understood. After being produced, a group of primary piRNAs associates Piwi proteins and directs them to repress transposons at the transcriptional level in the nucleus. Other than their direct role in repressing transposons, primary piRNAs can also initiate the production of secondary piRNA. piRNAs with such function are loaded in a second PIWI protein named Aubergine (Aub). Similar to Piwi, Aub is guided by piRNAs to identify its targets through base-pairing. Differently, Aub functions in the cytoplasm by cleaving transposon mRNAs. The 5' cleavage products are not degraded but loaded into the third PIWI protein Argonaute3 (Ago3). It is believed that an unidentified nuclease trims the 3' ends of those cleavage products to 23–29 nt, becoming mature piRNAs remained in Ago3. Such piRNAs whose 5' ends are generated by another PIWI protein are named secondary piRNAs. Intriguingly, secondary piRNAs loaded into Ago3 also cleave transposon mRNA or piRNA cluster transcripts and produce more secondary piRNAs loaded into Aub. This reciprocal feed-forward loop, named the “Ping-Pong cycle”, amplified piRNA abundance. By dissecting and analyzing data from large-scale deep sequencing of piRNAs and transposon transcripts, my dissertation research elucidates the biogenesis of germline piRNAs in Drosophila. How primary piRNAs are processed into mature piRNAs remains enigmatic. I discover that primary piRNA signal on the genome display a fixed periodicity of ~26 nt. Such phasing depends on Zucchini, Armitage and some other primary piRNA pathway components. Further analysis suggests that secondary piRNAs bound to Ago3 can initiate phased primary piRNA production from cleaved transposon RNAs. The first ~26 nt becomes a secondary piRNA that bind Aub while the subsequent piRNAs bind Piwi, allowing piRNAs to spread beyond the site of RNA cleavage. This discovery adds sequence diversity to the piRNA pool, allowing adaptation to changes in transposon sequence. We further find that most Piwi-associated piRNAs are generated from the cleavage products of Ago3, instead of being processed from piRNA cluster transcripts as the previous model suggests. The cardinal function of Ago3 is to produce antisense piRNAs that direct transcriptional silencing by Piwi, rather to make piRNAs that guide post-transcriptional silencing by Aub. Although Ago3 slicing is required to efficiently trigger phased piRNA production, an alternative, slicing-independent pathway suffices to generate Piwi-bound piRNAs that repress transcription of a subset of transposon families. The alternative pathway may help flies silence newly acquired transposons for which they lack extensively complementary piRNAs. The Ping-Pong model depicts that first ten nucleotides of Aub-bound piRNAs are complementary to the first ten nt of Ago3-bound piRNAs. Supporting this view, piRNAs bound to Aub typically begin with Uridine (1U), while piRNAs bound to Ago3 often have adenine at position 10 (10A). Furthermore, the majority of Ping-Pong piRNAs form this 1U:10A pair. The Ping-Pong model proposes that the 10A is a consequence of 1U. By statistically quantifying those target piRNAs not paired to g1U, we discover that 10A is not directly caused by 1U. Instead, fly Aub as well as its homologs, Siwi in silkmoth and MILI in mice, have an intrinsic preference for adenine at the t1 position of their target RNAs. On the other hand, this t1A (and g10A after loading) piRNA directly give rise to 1U piRNA in the next Ping-Pong cycle, maximizing the affinity between piRNAs and PIWI proteins.

Indiana University-Purdue University Indianapolis (IUPUI) / RNA-binding proteins (RBPs) are vital post-transcriptional regulatory molecules in transcriptome of mammalian species. It necessitates studying their expression dynamics to extract how post-transcriptional networks work in various mammalian tissues. RNA binding proteins (RBPs) play important roles in controlling the post-transcriptional fate of RNA molecules, yet their evolutionary dynamics remains largely unknown. As expression profiles of genes encoding for RBPs can yield insights about their evolutionary trajectories on the post-transcriptional regulatory networks across species, we performed a comparative analyses of RBP expression profiles across 8 tissues (brain, cerebellum, heart, lung, liver, lung, skeletal muscle, testis) in 11 mammals (human, chimpanzee, gorilla, orangutan, macaque, rat, mouse, platypus, opossum, cow) and chicken & frog (evolutionary outgroups). Noticeably, orthologous gene expression profiles suggest a significantly higher expression level for RBPs than their non-RBP gene counterparts, which include other protein-coding and non-coding genes, across all the mammalian tissues studied here. This trend is significant irrespective of the tissue and species being compared, though RBP gene expression distribution patterns were found to be generally diverse in nature. Our analysis also shows that RBPs are expressed at a significantly lower level in human and mouse tissues compared to their expression levels in equivalent tissues in other mammals: chimpanzee, orangutan, rat, etc., which are all likely exposed to diverse natural habitats and ecological settings compared to more stable ecological environment humans and mice might have been exposed, thus reducing the need for complex and extensive post-transcriptional control. Further analysis of the similarity of orthologous RBP expression profiles between all pairs of tissue-mammal combinations clearly showed the grouping of RBP expression profiles across tissues in a given mammal, in contrast to the clustering of expression profiles for non-RBPs, which frequently grouped equivalent tissues across diverse mammalian species together, suggesting a significant evolution of RBPs expression after speciation events. Calculation of species specificity indices (SSIs) for RBPs across various tissues, to identify those that exhibited restricted expression to few mammals, revealed that about 30% of the RBPs are species-specific in at least one tissue studied here, with lung, liver, kidney & testis exhibiting a significantly higher proportion of species specifically expressed RBPs. We conducted a differential expression analysis of RBPs in human, mouse and chicken tissues to study the evolution of expression levels in recently evolved species (i.e., humans and mice) than evolutionarily-distant species (i.e., chickens). We identified more than 50% of the orthologous RBPs to be differentially expressed in at least one tissue, compared between human and mouse, but not so between human and an outgroup chicken, in which RBP expression levels are relatively conserved. Among the studied tissues (brain, liver and kidney) showed a higher fraction of differentially expressed RBPs, which may suggest hyper- regulatory activities by RBPs in these tissues with species evolution. Overall, this study forms a foundation for understanding the evolution of expression levels of RBPs in mammals, facilitating a snapshot of the wiring patterns of post-transcriptional regulatory networks in mammalian genomes. In our second study, we focused on elucidating novel features of post-transcriptional regulatory molecules called as circRNA from LongPolyA RNA-sequence data. The debate over presence of nonlinear exon splicing such as exon-shuffling or formation of circularized forms has finally come to an end as numerous repertoires have shown of their occurrence and presence through transcriptomic analyses. It is evident from previous studies that along with consensus-site splicing non-consensus site splicing is robustly occurring in the cell. Also, in spite of applying different high-throughput approaches (both computational and experimental) to determine their abundance, the signal is consistent and strongly conforming the plausible circularization mechanisms. Earlier studies hypothesized and hence focused on the ribo-minus non-polyA RNA-sequence data to identify circular RNA structures in cell and compared their abundance levels with their linear counterparts. Thus far, the studies show their conserved nature across tissues and species also that they are not translated and preferentially are without poly (A) tail, with one to five exons long. Much of this initial work has been performed using non-polyA sequencing thus probably underestimates the abundance of circular RNAs originating from long poly (A) RNA isoforms. Our hypothesis is if the circular RNA events are not the artifact of random events, but has a structured and defined mechanism for their formation, then there would not be biases on preferential selection / leaving of polyA tails, while forming the circularized isoforms. We have applied an existing computational pipeline from earlier studies by Memczack et. al., on ENCODE cell-lines long poly (A) RNA-sequence data. With the same pipeline, we achieve a significant number of circular RNA isoforms in the data, some of which are overlapping with known circular RNA isoforms from the literature. We identified an approach and worked upon to identify the precise structure of circular RNA, which is not plausible from the existing computational approaches. We aim to study their expression profiles in normal and cancer cell-lines, and see if there exists any pattern and functional significance based on their abundance levels in the cell.

In recent biomedical research, graph representation learning and cell profiling techniques have emerged as transformative tools for analyzing high-dimensional biological data. The integration of these methods, as investigated in this study, has facilitated an enhanced understanding of complex biological systems, consequently improving drug discovery. The research aimed to decipher connections between chemical structures and cellular phenotypes while incorporating other biological information like proteins and pathways into the workflow. To achieve this, machine learning models' efficacy was examined for classification and regression tasks. The newly proposed graph-level and bio-graph integrative predictors were compared with traditional models. Results demonstrated their potential, particularly in classification tasks. Moreover, the topology of the COVID-19 BioGraph was analyzed, revealing the complex interconnections between chemicals, proteins, and biological pathways. By combining network analysis, graph representation learning, and statistical methods, the study was able to predict active chemical combinations within inactive compounds, thereby exhibiting significant potential for further investigations. Graph-based generative models were also used for molecule generation opening up further research avenues in finding lead compounds. In conclusion, this study underlines the potential of combining graph representation learning and cell profiling techniques in advancing biomedical research in drug repurposing and drug combination. This integration provides a better understanding of complex biological systems, assists in identifying therapeutic targets, and contributes to optimizing molecule generation for drug discovery. Future investigations should optimize these models and validate the drug combination discovery approach. As these techniques continue to evolve, they hold the potential to significantly impact the future of drug screening, drug repurposing, and drug combinations.

Följande studie behandlar utvecklandet av automatiserade cellsegmenteringsmodeller med avsikt att identifiera gränser mellan celler i hjärtvävnad. Syftet är att möjliggöra analys av data genererad från multiplexed error-robust in situ hybridization (MERFISH). MERFISH är en spatial transcriptomics-teknik som till skillnad från exempelvis single-cell RNA sequencing (ScRNA-seq) och single molecule fluorescence in situ hybridization (smFISH), möjliggör profilering av hundratals RNA-sekvenser hos enskilda celler utan att förlora dess rumsliga kontext. I Kosuri laboratoriet på Salk Institute of Biological Studies i San Diego tillämpas MERFISH på mushjärtan. Syftet är att få en djupare insikt i hur celler är organiserade i friska hjärtan, och hur denna struktur ändras i och med åldring och sjukdom. Att extrahera meningsfull information från MERFISH medför dock en betydande utmaning - en exakt cellsegmentering. Studien bidrar följaktligen till utvecklandet av segmenteringsmodeller för att kringgå de utmaningar som står i vägen för all efterföljande analys. Då klassiska segmenteringsalgoritmer är otillräckliga för att segmentera den komplexa vävnad som hjärtat utgörs av, tillämpades några av dagens mest avancerade och framstående maskininlärningsalgoritmer inom fältet, kallade Cellpose och Omnipose. Givet den täta och heterogena hjärtvävnaden, som härstammar från en bred distribution av celltyper och geometrier, utvecklades två separata modeller; en för att täcka både mindre celler och kardiomyocyter skurna på tvärsnittet; och en för att enbart segmentera kardiomyocyter skurna i longitudinell riktning. Den förstnämnda modellen utvecklades och tränades i Cellpose, och uppnådde en träffsäkerhet på 91.2%. Modellen för longitudinella kardiomyocyter utvecklades istället både i Cellpose och Omnipose för att utvärdera vilket nätverk som är bäst lämpat för ändamålet. Ingen av nätverken lyckades uppnå en tillräckligt hög träffsäkerhet för att vara applicerbar, och är därmed i behov av fortsatt träning. Modellen genererad i Omnipose bedöms dock vara mest lovande, givet dess mer heltäckande segmentering. Ytterligare utvecklingsområden för framtiden innefattar segmentering av celler i fibros-täta regioner, samt att utveckla en 3D-segmentering av hela hjärtat för att uppnå en mer komplett MERFISH-analys. Sammanfattningsvis har de genererade segmenteringsmodellerna banat väg för möjliggörandet av en rigorös MERFISH-analys av hjärtat. Genom att avslöja några av de strukturella och funktionella orsakerna till hjärtsvikt på en cellulär nivå, kan vi således på sikt bidra till utvecklingen av mer effektiva terapeutiska strategier. / The following study delves into the development of automated cell segmentation models, with the intention of identifying boundaries between cells in the cardiac tissue for analysing spatial transcriptomics data. Addressing the limitations of alternative techniques like single-cell RNA sequencing (ScRNA-seq) and single molecule fluorescence in situ hybridization (smFISH), the study underscores the innovative use of multiplexed error-robust fluorescence in situ hybridization (MERFISH) deployed by the Kosuri Lab at Salk Institute for Biological Studies. This advanced imaging-based technique allows for a single-cell transcriptome profiling of hundreds of different transcripts while retaining the spatial context of the tissue. The technique can accordingly reveal how the organization of cells within a healthy heart is altered during disease. However, the extraction of meaningful data from MERFISH poses a significant challenge - accurate cell segmentation. This thesis therefore presents the development of a robust model for cell boundary identification within cardiac tissue, leveraging some of the advanced supervised machine learning algorithms in the field, named Cellpose and Omnipose. Due to the dense and highly heterogeneous tissue- stemming from a wide distribution of cell types and shapes- two separate models had to be developed; one that covers the smaller cells and the cross-sectioned cardiomyocytes, and correspondingly one to cover the longitudinal cardiomyocytes. The cross-section model was successfully developed to achieve an accuracy of 91.2%, whereas the longitudinal model still needs further improvements before being implemented. The thesis acknowledges potential areas for improvement, emphasizing the need to further improve the segmentation of longitudinal cardiomyocytes, tackle the challenges with segmenting cells within fibrotic regions of the diseased heart, as well as achieving a precise 3D cell segmentation. Nonetheless, the generated models have paved the way towards enabling efficient downstream MERFISH analysis to ultimately understand the structural and functional dynamics of heart failure at a cellular level, aiding the development of more effective therapeutic strategies.