Advances in multiplex fluorescence imaging now allow the measurement of more than 50protein markers in whole tissue sections at single-cell resolution. This promises to reveal tumor biology at an unprecedented level of detail, both in undisturbed growth and in therapy. However, to quantitatively analyze these images, the images must be broken down into the basic units of tumor biology: single cells and their types. In this study, we applied a graph-based unsupervised clustering method, Leiden, to perform cell type identification in highly multiplexed fluorescence images, and based on the annotated images, we ran the tumor microenvironment niches analysis in order to resolve the recurring patterns of tumor microarchitecture. This thesis first introduces several potentially feasible clustering methods selected based on the structure of the datasets studied. The performance and stability of these clustering methods were compared. The project involved benchmarking different dimensionality reduction and clustering techniques on manually annotated reference datasets and healthy tissue with known cellular composition. It was ultimately determined that appropriate data transformations combined with Leiden clustering methods with proper parameters could automatically identify cells in a way coherent with established marker profiles. The results imply that Leiden clustering can also identify clusters of cells with novel marker combinations. Careful examination of the multiplex images shows that the markers are indeed found in the tumor, leading to new hypotheses regarding tumor biology. Tumor microenvironment niches analysis found several archetypal niches with specific cellular composition, indicating active accumulation of immune cells after radiotherapy, and the less vascularized feature of rebound glioblastomas after treatment. We hope to further validate our analysis to provide new insights into the pathological process of glioblastoma. In future research, the analysis pipeline is planned to be improved so that it can be robustly used to analyze the growing data of multiplexed tumor images, both in mouse cancer models or patient samples.

In this thesis, two algorithms that detect cross-contamination in tumor samples sequenced with next-generation sequencing (NGS) were evaluated. In genomic medicine, NGS is commonly used to sequence tumor DNA to detect disease-associated genetic variants and determine the most suitable treatment option. Targeted NGS panels are often employed to screen for genetic variations in a selection of specific tumor-associated genes. NGS handles samples from multiple patients in parallel, which poses a risk of cross-contamination between samples. Contamination is a significant issue in the interpretation of NGS results, as it can lead to the incorrect identification of genetic variants and, consequently, incorrect treatment. Therefore, contamination detection is a crucial quality control steps in the analysis of NGS data. Numerous algorithms for detection of cross-contamination have been developed, but many of these algorithms are not suited for small, targeted NGS panels, and several are not developed for tumor data. In this thesis, GATK's CalculateContamination and a self-created algorithm called ContaCheck were evaluated on simulated tumor NGS data. NGS samples were generated in silico with a Python script called BAMSynth and mixed to simulate cross-contamination with contamination rates between 1% and 50%. ContaCheck accurately detected contaminations ranging from 3% to 50% and identified the correct contaminant with an accuracy of 94%. CalculateContamination, on the other hand, detected contaminations ranging from 1% to 15% relatively accurately, but consistently failed to detect high level contaminations. The study showed that ContaCheck outperformed CalculateContamination on simulated NGS data, but to determine which algorithm is the best on real data and determine ContaCheck's applicability in a clinical setting, the algorithms need to be further evaluated on real tumor NGS samples.

Symbiodiniaceae is a family of dinoflagellates which often live in a symbiotic relationship with cnidarian hosts such as corals. Symbiodiniaceae are vital for host survival, providing energy from photosynthesis and in return gaining protection from environmental stress and nutrients. However, when these symbiont cells are exposed to environmental stress such as elevated temperatures they can be expelled from their host, leading to the coral bleaching, a global issue. Coral reefs are vital for marine biodiversity and hold a large economic importance due to fishing and tourism.  This thesis aims to develop a computational pipeline to study growth, shape and size of Symbiodiniaceae cells, which takes microscopy images using a mother machine microfluidics device and segments the Symbiodiniaceae cells. This enables extraction ofcellular features such as area, circularity and cell count to study morphology and growth of Symbiodiniaceae based on segmentation labels. To achieve this, pretrained segmentation models from the Cellpose algorithm were evaluated to decide which was the best to use to extract features most accurately. The results showed the pretrained ‘cyto3’ model with default parameters performed the best based on the Dice score. The feature extraction showed indications of division events of Symbiodiniaceae linked to light and dark cycles, suggesting synchronicity among cells. However, segmentation needs further investigation to accurately capture cells and add statistical significance to the feature extraction.

It is currently understood that diseases are typically not caused by rogue errors in genetics but have both molecular and environmental causes from myriad overlapping interactions within an interactome. Genetic errors, such as that seen by a single-nucleotide polymorphism can lead to a dysfunctional cell, which in turn can lead to systemic disruptions that result in disease phenotypes. Perturbations within the interactome, as can be caused by many such errors, can be organized into a pathophenotype, or “disease module”. Disease modules are sets of correlated variables that can represent many of a disease’s activities with subgraphs of nodes and edges. Many methods for inferring disease modules are available today, but the results each one yields is not only variable between methods but also across datasets and trial attempts. In this study, several such inference methods for deriving disease modules are evaluated by combining them to create “consensus” modules. The method of focus is Double-Specific Betweenness (S2B), which uses betweenness centrality across separate diseases to derive new modules. This study, however, uses S2B to combine the results of independent inference methods rather than separate diseases to derive new modules. Pre-processed asthma and arthritis data are compared using various combinations of inference methods. The performance of each result is validated using Pathway Scoring Algorithm. The results of this study suggest that combining methods of inference using MODifieR or S2B may be beneficial for deriving meaningful disease modules.

Transposons are selfish, mobile genetic elements, moving within the genome. The transposase genemakes this possible, as it codes for the enzyme that catalyzes the movement. In the case of bacteria,they can also move horizontally between individual bacteria, and sometimes even between species.By default, they are a burden for the host organism, coding for a protein that the host does not need.They also pose the risk of disabling the host’s crucial genes by inserting themselves into it.Transposons are under some pressure to benefit the host, to help propagate themselves moreeffectively. And some transposons have indeed evolved to benefit the host. Lactobacillus kunkeei is a bacterial species known to reside in honeybee guts. It is known for itsrole in honey preservation and wine spoilage. The genome of L. kunkeei is reduced because it is asymbiont, however it contains an unusually high amount of transposons in its genome. In this study, the transposase genes (transposon enzymes) found in L. kunkeei are studied andcategorized. The L. kunkeei have been extracted from honeybees (Apis mellifera). The honeybeesthemselves have been collected from the islands Åland and Gotland. This study focuses on the transposase genes that come in pairs, one after another in the genome.Transposase genes were identified using annotation software and orthology-based methods. Theannotation software provides numbering for the genes, which allows finding paired genes. Thepaired genes were categorized based on alignments and phylogenetic software. Pseudogenizedtransposons were identified based on length and/or clustering into triplets. A total of 766 paired transposase genes were found. The transposase genes were found to take up1.9% of the genome, on average. A low level of diversity has been found when performingalignments and generating phylogenetic trees. The positions of the transposase genes are generallyconserved within phylogenetic groups. Pseudogenization has been detected for some transposasegenes – 4.5 per genome, on average. All of the studied transposons belong to the IS3 family, whichis a family of Class I transposons.

Many diseases, such as cardiovascular diseases, cancer and diabetes, originate from several malfunctions in biological systems. The human body is regulated by a wide range of biological systems, composed of biological entities interacting in complex networks, responsible for carrying out specific functions. Some parts of the networks, such as hubs serving as master regulators, are more important for maintaining a function. To find the cause of diseases, where hubs are possible disease regulators, it is critical to know the structure of these biological systems. Such structures can be reverse engineered from high-throughput data with measured levels of biological entities. However, the complexity of biological systems makes inferring their structure a complicated task, demanding the use of computational methods, called network inference methods. Today, many network inference methods have been developed, that predicts the interactions of biological networks, with varying degree of success. In the DREAM5 challenge 35 network inference methods were evaluated on how well interactions in gene regulatory networks (GRNs) were predicted. Herein, in contrast to the DREAM5 challenge, we have evaluated network inference methods’ ability to predict hubs in GRNs. In accordance with the DREAM5 challenge, different methods performed the best on different data sets. Moreover, we discovered that network inference methods were not able to identify hubs from groups of similarly expressed genes. Also, we noticed that hubs in GRNs had a distinct expression in the data, leading to the development of a new method (the PCA method) for the prediction of hubs. Furthermore, the DREAM5 challenge showed that community predictions, combining the predictions from many network inference methods, resulted in more robust predictions of interactions. Herein, the community approach was applied on predicting hubs, with the conclusion that community predictions is the more robust approach. However, we also concluded that it was enough to combine 6-7 network inference methods to achieve robust predictions of hubs.

Bacterioplankton provide important ecosystem functions by carrying out biogeochemical cycling of organic matter. Playing an important role in the microbial loop they help remineralize carbon and nutrients. Bacteria also interact with phytoplankton during phytoplankton blooms. However, fundamental understanding on the underlying molecular mechanisms involved in the degradation of phytoplankton-derived organic matter is still in its infancy. Therefore, we analysed data from a mesocosm experiment following a natural phytoplankton-bloom from an upwelling system in the North- East Atlantic Ocean. The purpose was to contribute a mechanistic understanding based on functional gene expression analysis of natural microbial assemblages. Our results show the difference in functional gene expression within a bacterial metacommunity and how this functional response drastically switches between bloom build up and senescence. Transcripts showed a broad change in gene expression involving major SEED categories, with the bloom senescence phase exhibiting a higher relative abundance in major categories such as Carbohydrates, Protein Metabolism and Amino Acids and Derivatives. Within these categories genes connected to carbon utilization and transport systems (Ton and Tol) as well as chemotaxis showed a higher abundance during bloom senescence. The change in functionality based on transcripts showed a different bacterial community composition appearing over a very short time. We thus conclude that the bacterial functional gene expression response between build-up and degradation bloom phases is remarkably different and associated with a change in the identity of bacteria with active expression. Our findings highlight the importance of bacterial substrate specialists with different functional roles during different time points of phytoplankton blooms.

Using computational methods to identify protein-protein interactions (PPIs) supports experimental techniques by using less time and less resources. Identifying PPIs can be made through a template-based approach that describes how unstudied proteins interact by aligning a common structural template that exists in both interacting proteins. A pipeline that uses this is InterPred, that combines homology modelling and massive template comparison to construct coarse interaction models. These models are reviewed by a machine learning classifier that classifies models that shows traits of being true, which can be further refined with a docking technique. However, InterPred is dependent on using complex structural information, that might not be available from unstudied proteins, while it is suggested that PPIs are dependent of the shape and interface of proteins. A method that aligns structures based on the interface attributes is InterComp, which uses topological and sequence-order independent structural comparison. Implementing this method into InterPred will lead to restricting structural information to the interface of proteins, which could lead to discovery of undetected PPI models. The result showed that the modified pipeline was not comparable based on the receiver operating characteristic (ROC) performance. However, the modified pipeline could identify new potential PPIs that were undetected by InterPred.

The composition and diversity of the gastrointestinal microbiota and its interaction with human cells have been frequently associated with immune system functions and disease development, including autoimmunity and allergy. This is believed to be mediated in part through epigenetic modifications, mainly as DNA methylation. Several studies have collectively supported the beneficial effects of probiotics for the prevention of allergic disease. However, there have been few studies addressing the possibility for probiotic supplementation to induce epigenetic changes and its importance for allergy development. This study aims to investigate whether probiotic treatment with Lactobacillus reuteri, distributed during and after the pregnancy period, leads to epigenetic changes in the offspring and if this have any effect on the development of allergic disease. DNA methylation data received from a clinical allergy prevention study was analysed through a set of bioinformatics methods and basic network analysis. The obtained results suggests that supplementation with L. reuteri indeed induces some significant changes in DNA methylation. These changes did not exhibit any significant correlation with allergy outcome of the children. Furthermore, the methylation changes were found at positions located in genes not enriched for any allergy-related biological pathways. However, when taking the genes interactions with other genes into account an interconnected gene interaction module could be identified that showed enrichment for biological processes involved in the T cell receptor signaling pathway, central for immune response transduction. Further analyses did not fit into the time-frame of this thesis, but the obtained results gives a first informative view of the effects of L. reuteri on methylation patterns, and points out directions for the continuing project work.

Identifying homologous genes, that is genes from a common ancestor, is important in comparative genomic studies for understanding gene annotation and the predicted function of a gene. Several pieces of software, of which the most well-known is BLAST, have been developed for identifying homologues, but this can be challenging in non-model organisms where sometimes poor quality of genome assemblies and lack of annotation make it difficult to robustly identify homologues. The aim of this project was to build a bioinformatic framework for homology detection using genomes from non-model organisms. The approach developed used genome annotations, annotated polypeptide sequences and genome assembly sequences to detect homologous genes.The framework was applied to identify Drosophila melanogaster homologous wing development genes in the genomes of nine other insect species with the aim to understand the evolution of loss of wings. To identify changes related to wing loss, the homologous protein sequences obtained were aligned and phylogenetic trees were built from them. The aim of creating the multiple protein alignments and phylogenetic trees was to shed light on whether changes in gene sequences can be related to presence or absence of wings. From the set of 21 candidate wing development genes identified with literature and subsequent database searches, I tested eight and was successful in identifying homologues for all of them in eight of the 10 in sectgenomes. This was done using a combination of text searches in genome annotations, searches with Exonerate v. 2.4.0 alignment program in annotated polypeptide sequences and in genome assemblies. The eight genes chosen for testing the framework were based on initial finding of putative homologues in the eight insect genomes when using the first two steps of the framework. For the set of homologous wing development genes examined I was not able to identify any conclusive pattern of potential protein coding changes that correlated with loss of wings in these species. Improvement to the current pipeline could include using query sequences from closer relatives of the 8 test species than D. melanogaster and, of course, testing of the remaining wing development genes as well as further literature study of wing development genes. Together these could improve future studies on the evolution of wing loss in insects.