Motivation. DNA molecules mutate thousands of times every day. Some mutations are harmful to human cells, and may lead to the loss of function in important genes involved in DNA damage repair (DDR) mechanisms. Diseases such as tumors can exploit mutations in important, driver DDR genes to rapidly proliferate. Specific patterns of mutations (or signatures) are insightful indicators for the presence of DDR malfunctioning, which can be exploited to provide targeted treatment (e.g., by leveraging synthetic lethalities). Different methods have been developed to successfully extract relevant mutational signatures from the genomes of tumor patients. Most approaches are unsupervised and thus do not optimize toward distinguishing DDR deficiencies (DDRd). Supervised approaches achieve this, but rely on labeled in vitro data from tumor cell line genomes during training, due to the lack of DDRd ground truth for tumor patient genomes. Semi-supervised learning could bridge the gap and jointly exploit labeled cell line and unlabeled patient mutation profiles to generalize to patient tumors and provide more clinically relevant DDRd mutational signatures.
Results. We propose Pseudo-labeling Semi-Supervised NMF (PSS-NMF), a novel integrated signature extraction and label prediction method, which extends supervised non-negative matrix factorization (NMF) with the ability to incorporate unlabeled samples into the training via pseudo-labeling. Models learned using PSS-NMF were benchmarked on two different tasks, cancer type and DDRd prediction. PSS-NMF consistently improved prediction for patient tumors over the supervised NMF baseline for both tasks, learning signatures that better transferred to the patient tumor domain: the models achieved Macro F1 scores of 0.3842 and 0.1331 respectively for cancer type prediction, and 0.4928 vs 0.4704 for DDRd prediction. We further validated that PSS-NMF identified DDRd signatures were biologically relevant, by comparing them to known DDRd-related mutational signatures curated in COSMIC and investigating their exposures in patient tumor genomes.

Targeted and successful cellular therapies for disease treatment require an extensive mapping of the complex structure and dynamics of molecular mechanisms which determine the behaviour and function of cell. CELL-seq is a genome-wide screening procedure measuring specific and targeted protein quantities as phenotypic readouts and is employed by the Netherlands Cancer Institute to analyze which genes regulate the protein state of interest. This research aims to explore the current compendium of CELL-seq screens that investigate a range of phenotypes, to create a mapping of gene-gene associations that share similar phenotypic profiles and elucidate biology that is hard to uncover with more conventional screening techniques.

We perform exploratory research to investigate the ability of the screen compendium to show network structures that reflect known biological processes. We find that with stringent requirements on interactions the screen compendium shows enrichment for a wide range of biological processes and known protein-protein interactions. We further conclude that the experimental design biases network behaviour and needs to be accounted for when constructing networks. We recommended a mutual k-nearest neighbor network construction approach, which yielded networks with the most biological relevance.
We compare the CELL-seq screens using findings from the approaches to the DepMap dataset, a well-known collection of synthetic lethality CRISPR screens, and find that the behaviour of these datasets is in many ways mirrored. We conclude that this is both due to the biology they represent and the differences in the number of screens in each dataset. Finally, we compare the coverage of biological processes between the HAP1 compendium and DepMap, and show large overlap in their coverage. Nonetheless, the differences they do show leads us to bring forward two hypotheses for gene-gene interactions that score strongly uniquely in the CELL-seq networks which are biologically plausible but are not found in DepMap or curated literature, warranting future investigations.

All code pertaining to the methods and figures in this work are hosted on GitLab by the High Performance Computing Facility of the Netherlands Cancer Institute. As such the code can be viewed by supervisors, but further details could be shared upon request.

This paper tackles the problem of sample selection bias in machine learning, where the assumption of train and test sets being drawn from the same distribution is often violated. Existing solutions in domain adaptation, such as semi-supervised learning techniques, aim to correct this bias, but their ability to generalize to unseen test sets remains unexplored. To address this issue, specific semi-supervised methods (self-training and co-training) are trained on biased training sets and tested with an unbiased test set drawn from the same distribution. The results of this paper demonstrate that the semi-supervised methods consistently outperformed or matched the baseline models, with self-training exhibiting greater improvement. Through this study, a promising approach is presented to mitigate sample selection bias in machine learning.

Sample selection bias occurs when the selected samples in a subset of the original data set follow a different distribution than the samples from the original data set. This type of bias in the training set could result in a classifier being unable to predict samples from a testing data set optimally. Domain adaptation techniques try to adapt classifiers to a possible bias in the training or testing set. Subspace mapping techniques specifically do this by trying to find common subspaces between the source and target domain, where the source domain is the domain with all samples used for training, and the target domain is the domain with samples that must be predicted. This project aims to evaluate the effectiveness of two subspace mapping techniques in mitigating sample selection bias. This research assumes that no data samples from a target domain are available, but only unlabelled samples coming from an underlying global domain. The two subspace mapping techniques that will be tested in this paper are subspace alignment (SA) and transfer component analysis (TCA). This paper will show that the subspace alignment method is more effective on data sets with fewer features and where the source and target domains are further away from each other. The transfer component analysis method is more effective when more training samples are available on data sets with fewer features and where the distance between the source and target domain is not too big. The effectiveness of both methods also depends on the type and form of the data sets they are used on.

Importance weighting is a class of domain adaptation techniques for machine learning, which aims to correct the discrepancy in distribution between the train and test datasets, often caused by sample selection bias. In doing so, it frequently uses unlabeled data from the test set. However, this approach has certain drawbacks: it requires retraining for each new test set and fails when the number of test samples is very small. Therefore, we seek to study the performance of importance weighting techniques when the unlabeled data comes from an underlying domain, instead of one specific test set. We propose an evaluation framework inspired from scenarios traditionally known for posing difficulties to importance weighting and apply it to two popular algorithms, KMM and KLIEP. Our results reveal that both algorithms produce statistically significant classification improvements in most experiments. However, their performance is highly dependent on the characteristics of the dataset and the sampling bias. In particular, class overlap seems to influence adaptation ability in the case of unequal conditional probabilities of the source and target domains, while the "intensity" of the sampling bias is an important confounding factor when the train set size is small.

Sample selection bias is a well-known problem in machine learning, where the source and target data distributions differ, leading to biased predictions and difficulties in generalization. This bias presents significant challenges for modern machine learning algorithms. To tackle this problem, researchers have focused on developing domain adaptation techniques that aim to create robust methods against sample selection bias. One approach is the use of minimax estimation techniques, which belong to the category of inference-based techniques. Despite the extensive research in developing these domain adaptation methods, there remains a critical need to evaluate their performance. This thesis explores the performance differences of various minimax estimation techniques in the presence of sample selection bias, providing insights into their effectiveness in mitigating the challenges posed by biased data. By understanding and evaluating the performance of these techniques, this research contributes to the advancement of domain adaptation methods and their application in real-world machine learning scenarios.

Domain adaptation allows machine learning models to perform well in a domain that is different from the available train data. This non-trivial task is approached in many ways and often relies on assumptions about the source (train) and target (test) domains. Unsupervised domain adaptation uses unlabeled target data to mitigate a shift or bias difference between the domains. Deep domain adaptation (DDA) is a powerful class of these methods, which utilizes deep learning to extract high-level features that are common across the domains and robust against the shift. These algorithms adapt to a specific target domain. This has two possible downsides. Firstly, the model might not generalize and thus require retraining for each new domain. Secondly, obtaining data for the target domain(s) might be difficult. There can be situations where both source and target domains originate from a “global” domain, from which the samples are selected in a biased way. We explore a new application of existing DDA methods and answer the question: How effective is deep domain adaptation when adapting with the global domain, instead of the target domain, in case of sample selection bias?. Results with synthetic data show that where target adaptation works, global adaptation also improves accuracy compared to supervised learning, although to a lesser extent.

Synthetic lethality (SL) is a relationship between two genes, exploited for targeted anti-cancer therapy, whereby functional loss of both genes induces cell death, but the functional loss of either gene alone is non-lethal. Computational prediction of SL gene pairs is sought after because it is expensive to do lab screening for SL. Existing SL labeled pairs from wet- lab experiments often focus on specific genes or pathways, resulting in notable selection bias. Current SL prediction methods ignore this bias when training on available SL labels, and fail to generalize if test sets follow a different selection bias. One way to mitigate bias is to incorporate unlabeled pairs during model learning. However, conventional semi-supervised methods such as self-training can reinforce bias by adding confidently pseudolabeled pairs, which tend to be most similar to previously included samples. We present DBST, a self-training strategy that addresses the issue by promoting diversity in the selection of pseudolabeled samples. This is achieved using metric learning to find a class-contrastive representation of the feature space, based on which DBST selects diverse (or dissimilar) pseudolabeled pairs. In results for five cancer types, semi-supervised models, including DBST, delivered improved SL prediction performance over the supervised model. Additionally, DBST successfully incorporated unlabeled samples that were more dissimilar among them compared to standard self-training. In experiments with differing biases between train and test sets, DBST showed a slight improvement in performance compared to the supervised model.

Double-strand break (DSB) repair is a critical cellular process which repairs breaks in both strands of the DNA double helix. Different repair mechanisms are tasked with repairing such breaks. Predicting deficiencies in repair mechanisms has been widely used for therapeutic purposes, such as targeting cancer cells that have specific DNA repair deficiencies. DSB repair, however, is not error-free, resulting in mutations. These mutations are also influenced by the DNA sequence surrounding the break site. To the best of our knowledge, sequence representations have not been considered when predicting DNA repair deficiencies. We hypothesise that higher-order information can be extracted from sequence representations. In this study, we research the problem of predicting Non-Homologous End Joining (NHEJ) repair deficiencies. Initially, we evaluate how accurately we can predict NHEJ repair deficiency using only the mutational outcome frequencies (mutational spectra). Afterwards, we examine how combining mutational spectra with representations of the sequence surrounding the break site can improve the prediction of NHEJ repair deficiency. We demonstrate that adding DNABERT sequence representations to mutational spectra features significantly improves prediction accuracy from 94.44% to 96.12%. We also show that even simple sequence representations, such as 1-mer frequencies, can lead to significant improvements. Our findings highlight the importance of including sequence representations with mutational spectra in repair deficiency prediction.

The inclusion of intronic reads in the downstream analysis of RNA-sequencing (RNA-seq) data has long been controversial. Recent studies show that intronic reads do contain relevant biological signal. Additionally, studies have discovered differential expression unique to intronic reads in certain diseases. Nevertheless, most disease prediction studies only use exonic read counts as input to their models. In this study, we investigate the informativeness of intronic read counts for RNA-seq-based machine learning prediction tasks. Furthermore, we explore possibilities to combine exonic and intronic read counts to increase predictive performance. To this end, we use an RNA-seq dataset originating from four different brain regions and try to predict multiple different clinical labels, including Alzheimer's disease and dementia. We start by identifying differently expressed genes by performing differential gene expression (DGE) analysis. Next, we evaluate the predictive performance of both exonic and intronic read counts using logistic regression. Subsequently, we explore some basic machine learning techniques to combine the information contained in both sets. Furthermore, we construct our own model architectures with the aim of gaining information by using both sets. We show, for this dataset, that exonic and intronic reads have overlapping but also unique differentially expressed genes. Using these genes we show that the predictive performance using the exonic and intronic reads is very similar for all predicted labels. We further show that even though different genes are identified, the biologically relevant signal for the prediction task appears to be the same in exonic and intronic read counts. We are not able to leverage the combination of the counts to further increase predictive performance. Existing disease prediction models have neglected the inclusion of intronic reads. In light of our findings, machine learning models that incorporate intronic reads could potentially discover novel biological insights.

Genomics has revolutionized our understanding of evolution, hereditary diseases, and more. The advent of long-read DNA sequencers i.e. Oxford Nanopore Technologies' innovations, has opened many new research potentials in genomics. These sequencers produce significantly longer DNA reads, facilitating novel applications. However, this technological leap brings challenges, particularly in accurate basecalling which is the process of converting raw sequenced measurements into digital base pair sequences. While advances in basecalling accuracy have been steadily improving over the years, the computational intensity remains a bottleneck in genomic analysis workflows, demanding costly high-end GPUs for probabilistic neural network models.

The main problem this thesis addresses is the implementation of an accelerated hardware solution for the compute-intensive process of basecalling long-read sequences. The thesis presents an FPGA-based implementation of the computationally demanding Long Short-Term Memory (LSTM) layers within the basecalling network known as Bonito. However, due to the lack of floating-point arithmetic units available on the FPGA, the FPGA implementation could not achieve competitive performance compared to GPUs.

While the FPGA implementation falls short of GPU performance, it serves as a possible stepping stone toward developing an ASIC solution for implementing the Bonito LSTM layers or potentially implementing the entire Bonito model. An ASIC implementation has the potential for superior performance up to 9 times faster than a GPU implementation while additionally being cost-effective. This suggests that ASICs hold promise as a future direction for accelerating long-read sequence basecalling, allowing for faster and more affordable genomics research.

Motivation: Many tumors show deficiencies in DNA damage repair. These deficiencies can play a role in the disease, but also expose vulnerabilities with therapeutic potential. Targeted treatments exploit specific repair deficiencies, for instance based on synthetic lethality. To decide which patients could benefit from such therapies requires the ability to determine the repair deficiency status of a tumor. It has been suggested that mutational signatures could be better predictors of DNA repair deficiency than loss of function in select genes. However, current models for prediction of repair deficiency rely on mutational signatures extracted using unsupervised learning techniques. As a result, the signatures are not optimized to discriminate between repair deficiency status or pathway. We argue that the supervised learning of mutational signatures guided by repair deficiency status could enable the identification of signatures that are predictive of repair deficiency, and capture underlying mechanisms of DNA repair.
Results: We propose S-NMF, a supervised non-negative matrix factorization method, which jointly optimizes two objectives: (1) learning of signatures shared across tumor samples using NMF, and (2) learning of signatures predictive of repair deficiency using logistic regression. We apply S-NMF to mutation profiles of human induced pluripotent cell lines carrying knockouts of genes involved in three DNA repair pathways: homologous recombination, base excision repair, and mismatch repair. We show that S-NMF achieves high prediction accuracy (0.971) and learns signatures that better distinguish the repair deficiency of a sample. Signatures extracted by S-NMF are similar to cancer-related signatures associated with the same repair deficiency. Additionally, S-NMF can capture signatures of deficiencies affecting distinct subpathways within a main repair pathway (e.g. OGG1 and UNG mechanisms in base excision repair).

Recent advancements in quantification of repair outcomes of CRISPR-Cas9 mediated double-stranded DNA breaks (DSBs) have allowed for the use of machine learning for predicting the frequencies of these repair outcomes. Local DNA sequence context influences the frequencies of mutations that arise when DNA gets repaired after it is targeted by CRISPR (CRISPR outcomes). Contemporary models exploit this and can predict what the frequencies are of CRISPR outcomes at predetermined genomic loci. Predictions of such models are reasonably precise, but there may be opportunities for improvement in how the DNA sequence context is leveraged for making predictions. Some models only utilize a set of hand-crafted features, limiting the available information for the model. Other models do utilize broader sequence context but disregard sequence order or only predict a limited set of outcome classes. In this work we present an attention-based deep learning model that uses DNA sequence context to make fine-grained CRISPR outcome predictions. We present a custom input embedding for representing DSB repair outcomes and we expand on existing methods for analyzing attention-based models.

Motivation: As one of the most common and life-threatening diseases in humans, cancer is a result of the accumulation of somatic mutations throughout the life cycle. Somatic mutation is a joint result of DNA lesion, which is a result of damage on DNA caused by mutagen, and the failure of DNA repair mechanisms. Therefore, understanding the mechanism of DNA repair is important for the development of cancer pathology. Research on mutational signatures in cancer genomic data has given us insights into the relationship between DNA repair pathway and certain types of cancer. However, close-up studies on DNA repair pathways still need to be carried out to understand repair behaviour further. The CRISPR-Cas9 technique can create double-strand break in DNA. This paper extracts mutational signature to observe the characteristic of the pathway for the CRISPR-mediated double-strand break repairing in order to obtain a deeper understanding on DNA repair mechanisms.
Results: By analyzing the mutational signature extracted from CRISPR-Cas9 mediated targets, we found out that the mutational processes involved in double-strand break DNA repair processes are simple and predictable compared to mutational processes involved in the formation of cancer cells. When comparing the mutational processes in wildtype and cNHEJ-deficient cells, the latter show a much higher frequency of micro-homology related deletions. However, no significant difference between the two types of cells on the mutational process is able to be captured by the mutational signature extraction algorithm. The experiments also show that the selection of mutational features could significantly influence the final signature extraction results.

Survival analysis is a statistical method used to predict when an event will occur. Machine learning survival models have been used in many cancer studies. However, machine learning models may not always be interpretable. The current lack of research for explainable survival analysis for urothelial cancer prompted this study. This study offers an insight into the generalizability and explainability of machine learning models for urothelial cancer. We also determine how we can make the models interpretable in the presence of collinearity. In this study, we compared the performance of the models; Rank Linear Support Vector Machine (SVM), Rank Kernel SVM, Coxnet, Random Survival Forest (RSF), and Gradient Boosting (Gboost). We used the Memorial Sloan Kettering (MSK) and The Cancer Genome Atlas (TCGA) datasets. We used gene expression variables and clinical variables to train our models. We evaluated these models based on the C-index. We used Permutation Feature Importance (PFI), a model-agnostic method, to explain our models and used Principal Component Analysis (PCA) to deal with collinearity. We determined that the best linear model was Rank Linear SVM (C-index = 0.58) and the best non-linear model was RSF (C-index = 0.63). Using PFI showed that some of the top-most important genes were expressed in urothelial cancer, one of them even being a prognostic marker. With PCA, we were able to deal with collinearity, and the performance using PCA was comparable to models not using it. PFI with PCA showed that processes exhibited in the top genes were prevalent in cancer.

Double strand breaks are lesions to the DNA and can be fatal for cells. Therefore these breaks are repaired, primarily by one of the three major repair pathways. Two of these pathways are non-homologous end-joining (NHEJ) and theta-mediated end-joining (TMEJ). These pathways leave genetic alterations in their repair products, a form of DNA damage. DNA damage is linked to several diseases such as cancer. Understanding of these pathways is important and being able to recognize which pathways are active can be beneficial for research. In this work, repair products are used to predict repair-deficient genotypes using Cas9-induced repair products. Ku80 and PolQ deficient genotypes are used, impairing NHEJ and TMEJ respectively. The ability to recognize a repair-deficient genotype is tested using two predictive tasks. First statistical machine learning algorithms are used to predict the genotype where a repair product can be found. This is done by only using a single repair product as input. Secondly, a set of Cas9-induced repair products from a single cell culture is used to predict the genotype of that cell culture. Results show that when given a single repair products, models have difficulty predicting the correct genotype. However, results are modest and the best classifier achieved an AUC of 0.76. For predicting the genotype of a cell culture using multiple repair products of that culture showed really promising results. When predicting on cell cultures with breaks induced on a target site which the model has seen in the training data, results are near perfect. Predicting on unseen target sites shows that there is room for improvement but the best performing models showed an average AUC of 0.879 across target sites. A Results show that Cas9-induced repair products can be used to predict repair-deficient genotypes.

Synthetic lethality (SL) arises between two genes when loss of function of both genes would lead cells to become inviable. This can be exploited for therapy, where a drug is used to selectively kill diseased cells by perturbing one gene of an SL pair where the other gene is inactive (e.g. through naturally occurring mutation). Computational prediction of SL relationships is very appealing as it can help reduce cost- and labour-intensive experimental testing to the most promising candidate pairs. Even though machine learning models have shown promising results for SL prediction compared to traditional statistical approaches, crucial questions remain. First, which sources of molecular data are most useful for SL prediction? Many approaches rely on either cell line or patient tumour data separately, and ignore data from healthy tissue. We argue these should be combined to leverage relevant data sources that are exclusively available for cancer cell models and patient tumours, and to enable the transfer of knowledge between models and actual patient tumours. Likewise, changes in the relationship of gene pairs between healthy and tumour tissue may be informative for SL prediction. We assess several machine learning techniques to best leverage molecular profiles for cancer-specific or pan-cancer SL prediction. Second, what are the effects of selection bias on SL prediction methods and which techniques are most robust? This has been insufficiently addressed, as models in the literature are often tested using data from one or two cancer types or datasets. We investigate robustness to cancer representation and gene selection biases, which are inherent to most SL datasets. We hypothesise that approaches based on matrix factorisation will be especially sensitive to the latter, as they are dependent on an a priori SL network structure, which also determines the scope of the prediction space.

Due to their altered genetic context, cancer cells can become dependent on specific genes for their survival. Such cancer-specific dependencies may represent promising therapeutic targets. However, knowledge on which molecular features of cancer cells induce specific dependencies is still limited and hampers the development of effective targeted therapies. Several large scale studies have systematically measured the dependency of hundreds of known cancer cell lines on thousands of genes using gene silencing. These data have enabled the learning of supervised models to predict dependencies of cancer cells on each gene based on molecular features of the cells. In particular, linear regression with regularization, such as Elastic Net, has been used to select molecular features associated with such dependencies. Since these approaches model dependencies for each gene independently, the selected features provide limited insight into common mechanisms underlying gene dependency. Moreover, they may fail to identify robust associations with gene dependency due to the small size of the available training data. In this work, we apply a multi-task learning approach (Macau) to learn the relationship between transcriptome and gene dependency in cancer cell lines for multiple genes simultaneously. To do so, Macau projects genes, cancer cell lines and their features into a shared latent space. We explore this latent space to go beyond linking individual transcriptomic features with dependencies, and further associate pathway changes with functionally related genes without enforcing prior knowledge on pathway structure. Although Macau and Elastic Net yield similar predictive performance, they find different kinds of associations. First, Macau favors features that are relevant for predicting dependency across multiple genes. Second, Macau captures inherent functional relationships between genes and leverages these to predict cancer gene dependencies. Additionally, Macau can recover similarities between cancer cell lines belonging to the same cancer type based on their dependencies only. In summary, modelling cancer dependencies simultaneously for multiple genes can reveal underlying mechanisms shared by functionally related genes, which would be missed when learning models independently per gene.