Characterizing genetic variation including point mutations and structural variations, is key to understanding phenotypic variation in humans. The rapid development of sequencing technology has fueled the development of computational methods for elucidating genetic variation. In this dissertation, we develop novel computational methods to mainly target two human genetic variation problems using current and emerging sequencing technology. Capturing variation on the haplotype level is challenging with current sequencing technology as it involves linking together short sequenced fragments of the genome that overlap at least two heterozygous sites. While there has been a lot of research on correcting errors to achieve accurate haplotypes, relatively little work has been done on designing sequencing experiments to get long haplotypes. With the development of new sequencing technology and experimental haplotyping methods, we parametrize the haplotyping problem in two contexts, strobe sequencing and clone-based haplotyping, and provide theoretical and empirical assessment of the impact of different parameters on haplotype length. Variation in certain regions of the genome are harder to capture than others. Reconstruction of the donor genome from whole genome sequence data is either based on de novo assembly of the short reads or on mapping reads to a standard reference genome. While these techniques work well for inferring 'simple' genomic regions, they are confounded by regions with complex variation patterns including regions of direct immunological relevance such as the HLA and KIR regions. Characterizing these regions have previously relied on laboratory methods using traditional and quantitative PCR primers and probes which can be labor and time intensive. We address the problem of ambiguous mapping in complex regions by defining a new scoring function for read-to-genome matchings. This scoring function is applied to predicted sequence assemblies of the KIR region in order to determine the most likely KIR haplotype groups of the donor. In another approach, we developing a novel method based on barcoding (deriving signatures) known KIR templates in order to determine the copy number and allelic type of genes in the KIR region directly from whole genome sequencing data without assembly or mapping

Abstract Background Humans are diploid, carrying two copies of each chromosome, one from each parent. Separating the paternal and maternal chromosomes is an important component of genetic analyses such as determining genetic association, inferring evolutionary scenarios, computing recombination rates, and detecting cis-regulatory events. As the pair of chromosomes are mostly identical to each other, linking together of alleles at heterozygous sites is sufficient to phase, or separate the two chromosomes. In Haplotype Assembly, the linking is done by sequenced fragments that overlap two heterozygous sites. While there has been a lot of research on correcting errors to achieve accurate haplotypes via assembly, relatively little work has been done on designing sequencing experiments to get long haplotypes. Here, we describe the different design parameters that can be adjusted with next generation and upcoming sequencing technologies, and study the impact of design choice on the length of the haplotype. Results We show that a number of parameters influence haplotype length, with the most significant one being the advance length (distance between two fragments of a clone). Given technologies like strobe sequencing that allow for large variations in advance lengths, we design and implement a simulated annealing algorithm to sample a large space of distributions over advance-lengths. Extensive simulations on individual genomic sequences suggest that a non-trivial distribution over advance lengths results a 1-2 order of magnitude improvement in median haplotype length. Conclusions Our results suggest that haplotyping of large, biologically important genomic regions is feasible with current technologies.

Abstract With the remarkable development in inexpensive sequencing technologies and supporting computational tools, we have the promise of medicine being personalized by knowledge of the individual genome. Current technologies provide high throughput, but short reads. Reconstruction of the donor genome is based either on de novo assembly of the (short) reads, or on mapping donor reads to a standard reference. While such techniques demonstrate high success rates for inferring 'simple' genomic segments, they are confounded by segments with complex duplication patterns, including regions of direct medical relevance, like the HLA and the KIR regions. In this work, we address this problem with a method for assessing the quality of a predicted genome sequence for complex regions of the genome. This method combines two natural types of evidence: sequence similarity of the mapped reads to the predicted donor genome, and distribution of reads across the predicted genome. We define a new scoring function for read-to-genome matchings, which penalizes for sequence dissimilarities and deviations from expected read location distribution, and present an efficient algorithm for finding matchings that minimize the penalty. The algorithm is based on a formal problem, first defined in this paper, called Coverage Sensitive many-to-many min-cost bipartite Matching (CSM). This new problem variant generalizes the standard (one-to-one) weighted bipartite matching problem, and can be solved using network flows. The resulting Java-based tool, called SAGE (Scoring function for Assembled GEnomes), is freely available upon request. We demonstrate over simulated data that SAGE can be used to infer correct haplotypes of the highly repetitive KIR region on the Human chromosome 19.

Abstract Background Haplotypes are important for assessing genealogy and disease susceptibility of individual genomes, but are difficult to obtain with routine sequencing approaches. Experimental haplotype reconstruction based on assembling fragments of individual chromosomes is promising, but with variable yields due to incompletely understood parameter choices. Results We parameterize the clone-based haplotyping problem in order to provide theoretical and empirical assessments of the impact of different parameters on haplotype assembly. We confirm the intuition that long clones help link together heterozygous variants and thus improve haplotype length. Furthermore, given the length of the clones, we address how to choose the other parameters, including number of pools, clone coverage and sequencing coverage, so as to maximize haplotype length. We model the problem theoretically and show empirically the benefits of using larger clones with moderate number of pools and sequencing coverage. In particular, using 140 kb BAC clones, we construct haplotypes for a personal genome and assemble haplotypes with N50 values greater than 2.6 Mb. These assembled haplotypes are longer and at least as accurate as haplotypes of existing clone-based strategies, whether in vivo or in vitro. Conclusions Our results provide practical guidelines for the development and design of clone-based methods to achieve long range, high-resolution and accurate haplotypes.

Objectives

To develop: (1) two validated risk prediction models for coronavirus disease-2019 (COVID-19) positivity using readily available parameters in a general hospital setting; (2) nomograms and probabilities to allow clinical utilisation.

Methods

Patients with and without COVID-19 were included from 4 Hong Kong hospitals. The database was randomly split into 2:1: for model development database (n = 895) and validation database (n = 435). Multivariable logistic regression was utilised for model creation and validated with the Hosmer-Lemeshow (H-L) test and calibration plot. Nomograms and probabilities set at 0.1, 0.2, 0.4 and 0.6 were calculated to determine sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV).

Results

A total of 1330 patients (mean age 58.2 ± 24.5 years; 50.7% males; 296 COVID-19 positive) were recruited. The first prediction model developed had age, total white blood cell count, chest x-ray appearances and contact history as significant predictors (AUC = 0.911 [CI = 0.880-0.941]). The second model developed has the same variables except contact history (AUC = 0.880 [CI = 0.844-0.916]). Both were externally validated on the H-L test (p = 0.781 and 0.155, respectively) and calibration plot. Models were converted to nomograms. Lower probabilities give higher sensitivity and NPV; higher probabilities give higher specificity and PPV.

Conclusion

Two simple-to-use validated nomograms were developed with excellent AUCs based on readily available parameters and can be considered for clinical utilisation.

Background

COPD increases susceptibility to sleep disturbances, which may in turn predispose to increased respiratory symptoms. The objective of this study was to evaluate, in a population-based sample, the relationship between subjective sleep quality and risk of COPD exacerbations.

Methods

Data were obtained from the Canadian Cohort Obstructive Lung Disease (CanCOLD) study. Participants with COPD who had completed 18 months of follow-up were included. Sleep quality was measured with the Pittsburgh Sleep Quality Index (PSQI) and a three-factor analysis. Symptom-based (dyspnea or sputum change ≥ 48 h) and event-based (symptoms plus medication or unscheduled health services use) exacerbations were assessed. Association of PSQI with exacerbation rate was assessed by using negative binomial regression. Exacerbation-free survival was also assessed.

Results

A total of 480 participants with COPD were studied, including 185 with one or more exacerbations during follow-up and 203 with poor baseline sleep quality (PSQI score > 5). Participants with subsequent symptom-based exacerbations had higher median baseline PSQI scores than those without (6.0 [interquartile range, 3.0-8.0] vs 5.0 [interquartile range, 2.0-7.0]; P = .01), and they were more likely to have baseline PSQI scores > 5 (50.3% vs 37.3%; P = .01). Higher PSQI scores were associated with increased symptom-based exacerbation risk (adjusted rate ratio, 1.09; 95% CI, 1.01-1.18; P = .02) and event-based exacerbation risk (adjusted rate ratio, 1.10; 95% CI, 1.00-1.21; P = .048). The association occurred mainly in those with undiagnosed COPD. Strongest associations were with Factor 3 (sleep disturbances and daytime dysfunction). Time to symptom-based exacerbation was shorter in participants with poor sleep quality (adjusted hazard ratio, 1.49; 95% CI, 1.09-2.03).

Conclusions

Higher baseline PSQI scores were associated with increased risk of COPD exacerbation over 18 months' prospective follow-up.