Student Number : 0213068F - MSc Dissertation - School of Computer Science - Faculty of Science / Drug resistance of the HI virus due to its fast replication and error-prone mutation is a key factor in the failure to combat the HIV epidemic. For this reason, performing pre-therapy drug resistance testing and administering appropriate drugs or combination of drugs accordingly is very useful. There are two approaches to HIV drug resistance testing: phenotypic (clinical) and genotypic (based on the particular virus’s DNA). Genotyping tests HIV drug resistance by detecting specific mutations known to confer drug resistance. It is cheaper and can be computerised. However, it requires being able to know or learn what mutations confer drug resistance. Previous research using pattern recognition techniques has been promising, but the performance needs to be improved. It is also important for techniques that can quickly learn new rules when faced with new mutations or drugs. A relatively recent addition to these techniques is the Support Vector Machines (SVMs). SVMs have proved very successful in many benchmark applications such as face recognition, text recognition, and have also performed well in many computational biology problems where the number of features targeted is large compared to the number of available samples. This paper explores the use of SVMs in predicting the drug resistance of an HIV strain extracted from a patient based on the genetic sequence of those parts of the viral DNA encoding for the two enzymes, Reverse Transcriptase or Protease, which are critical for the replication of the HIV virus. In particular, it is the aim of this reseach to design the model without incorporating the biological knowledge at hand to enable the resulting classifier accommodate new drugs and mutations. To evaluate the performance of SVMs we used cross validation technique to measure the unbiased estimate on 2045 data points. The accuracy of classification and the area under the receiver operating characteristics curve (AUC) was used as a performance measure. Furthermore, to compare the performance of our SVMs model we also developed other prediction models based on popular classification algorithms, namely neural networks, decision trees and logistic regressions. The results show that SVMs are a highly successful classifier and out-perform other techniques with performance ranging between (94.13%–96.33%) accuracy and (81.26% - 97.49%) AUC. Decision trees were rated second and logistic regression performed the worst.

M.Sc. (Med.), Faculty of Health Sciences, University of the Witwatersrand, 2009. / Haemophilia A is a common X-linked recessive bleeding disorder, affecting about 1 in 5000 males worldwide. It is caused by a deficiency of functional coagulation Factor VIII (FVIII), resulting in prolonged or abnormal bleeding episodes. The severity of the disease is related to the level of functional FVIII in the plasma. The FVIII gene is a large gene, located at Xq28 with a complex genomic organisation. It contains 26 exons spanning 186kb of genomic DNA, and produces a 9kb transcript, resulting in a functional protein of 2332 amino acids. Over 900 mutations, which span a wide variety of categories, including rearrangements; complete or partial gene deletions; large insertions; duplications; frameshift mutations; splicing defects; nonsense and missense mutations, have been identified in the FVIII gene. Most mutations are rare or family specific, except for the intron 22 inversion mutation, which is reported to account for 45-50% of mutations in severe haemophilia A patients in most populations. A second inversion mutation, in intron 1, accounts for approximately 3.8% of haemophilia A patients in the UK. In South Africa, diagnostic mutation testing is currently only available for the intron 22 inversion mutation. Linked marker analysis is used to track high risk alleles in families where the disease-causing mutation is unknown. This study aims to evaluate an mRNA-based method to identify disease-causing mutations in South African haemophilia A patients and improve the diagnostic service. Blood samples from 120 patients were tested first for the intron 22 and then for intron 1 inversion mutations. Inversion negative patients were analysed further using mRNA. A mutation has been identified in 73.3% (88/120) of all patients. 30% (36/120) of patients had the intron 22 inversion, 2.5% (3/120) an intron 1 inversion and 40.8% (49/120) of patients had a mutation identified by mRNA analysis. A mutation was not identified in the remaining 26.7% (32/120) due to sample and technical difficulties. Of the 49 mutations identified through mRNA analysis, 28 patients (57.1%) have a point mutation (17 missense (34.7%), 9 nonsense (18.4%) and 2 splice-site mutations (4.1%)), 9 patients (18.4%) have a deletion and 7 patients (14.3%) have an insertion. Another 5 patients (10.2%) have a complex mutation (including patients where an exon deletion was detected on mRNA analysis, but no mutation was identified on DNA analysis). One mutation, c.3637insA, was found recurrently in 14% (6/43) of patients from the white population. This single base insertion results in a frameshift mutation with a premature stop codon at amino acid 1221 (only translating about half the normal FVIII protein). This common mutation, together with haplotype analysis, suggests a founder effect for this mutation. mRNA analysis of the FVIII gene is a novel technique in mutation detection for haemophilia A. It decreases the costs involved in sequencing the coding region and it offers improved mutation detection compared to DNA analysis. Diagnostic testing in South Africa should be extended from the current intron 22 inversion mutation to include DNA analysis for the intron 1 inversion and the founder mutation (c.3637insA) in white patients, followed by mRNA testing, starting with the analysis of the fragments spanning exon 14. mRNA analysis identifies an additional 55.7% of mutations compared to conventional diagnostic testing for the intron 22 inversion alone.

A dissertation submitted to the Faculty of Health Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Masters of Science in Medicine, 2014 / Many loci coding for xenobiotic metabolising enzymes, especially those involved in carcinogen metabolism, confer susceptibility to various types of cancers. These genes have been poorly investigated in sub-Saharan African populations, where the genetic variation that exists is relatively unknown. The primary objectives of the study are to determine the frequency variation among 15 loci in sub-Saharan Africans, the level of genetic diversity, and the genetic affinities among sub-Saharan Africans. Secondary, the study aims to evaluate the implication of these variants in disease susceptibility, especially cancer. The study population comprised of 1880 unrelated individuals from 18 sub-Saharan African populations. DNA samples were used to examine genetic variation for phase I metabolism genes CYP1A1, CYP1A2, CYP2A6, CYP2D6, CYP2E1, and phase II metabolism genes GSTM1, GSTT1, GSTP1 and NAT2. A single base extension (SBE) method was designed and used to genotype single nucleotide polymorphisms (SNP): CYP1A1*2A and *2C; CYP1A2*1C and *1F; CYP2A6*7 and *8; CYP2D6*3A (2549delA) and CYP2D6*4(1846G>A); CYP2E1*5B(PstI) and CYP2E1*5B(RsaI); GSTP1*Ile105Val and *Ala114Val; and NAT2*14A. To investigate the presence of null mutations GSTM1*0 and GSTT1*0 a previously reported multiplex PCR method was used. The distribution of mutations in the sample was interpreted and compared with data from literature respectively. Mutations CYP1A1*2A, CYP1A2*1C, CYP1A2*1F, CYP2A6*7, CYP2D6*4 and GSTP1*Ile105Val mutations was found in most sub-Saharan Africans, while CYP1A1*2C, CYP2A6*8, CYP2D6*3A and GSTP1*Ala114Val mutations were almost non-existent. Both GSTM1*0 and GSTT1*0 mutations were present in all populations, with GSTT1*0 most frequent. The distribution of NAT2*14A confirms previous reports of its exclusive existence in Africans. Hardy-Weinberg Equilibrium (HWE) and Tajima’s D statistic tests showed none of the mutations were under selection. The genetic affinities of sub-Saharans were analysed. Bantu-speakers were closely related with little correlation to their geographic locations. Khoisan-speakers were closely related, genetically most distinct and oldest among populations. Pygmies were similarly distinct from most populations and one of the oldest surviving populations. The data further supports previous reports that the Khwe are descendants of an east African pastoralist group. AMOVA analyses revealed language as a major confounder among sub-Saharans. Haplotypes were inferred to determine their distribution and to understand their significance in populations with respect to their functional relevance. The study has confirmed previous reports of genetic histories of these sub-Saharan African populations. In unravelling the distribution of these mutations, the study has added to the global picture of these mutations. In doing so, the data may add value to the design of future cancer studies and pharmacological studies. The study also highlights the importance of elucidating ancestral relations of populations, more specifically linguistic and anthropological relationships, and to include in the design of future clinical trials in Africa.

Welsh grammatical gender exhibits several unusual properties. This paper argues that these properties are necessarily connected. The argument is based on a series of corpus investigations using techniques from statistical natural language processing, specifically distinguishing properties that exhibit significant statistical patterns from those which can be used to make useable predictions. Specifically, it’s shown that the grammatical properties of Welsh gender are such that its unusual statistical properties follow.

abstract: Ideas from coding theory are employed to theoretically demonstrate the engineering of mutation-tolerant genes, genes that can sustain up to some arbitrarily chosen number of mutations and still express the originally intended protein. Attention is restricted to tolerating substitution mutations. Future advances in genomic engineering will make possible the ability to synthesize entire genomes from scratch. This presents an opportunity to embed desirable capabilities like mutation-tolerance, which will be useful in preventing cell deaths in organisms intended for research or industrial applications in highly mutagenic environments. In the extreme case, mutation-tolerant genes (mutols) can make organisms resistant to retroviral infections. An algebraic representation of the nucleotide bases is developed. This algebraic representation makes it possible to convert nucleotide sequences into algebraic sequences, apply mathematical ideas and convert results back into nucleotide terms. Using the algebra developed, a mapping is found from the naturally-occurring codons to an alternative set of codons which makes genes constructed from them mutation-tolerant, provided no more than one substitution mutation occurs per codon. The ideas discussed naturally extend to finding codons that can tolerate t arbitrarily chosen number of mutations per codon. Finally, random substitution events are simulated in both a wild-type green fluorescent protein (GFP) gene and its mutol variant and the amino acid sequence expressed from each post-mutation is compared with the amino acid sequence pre-mutation. This work assumes the existence of synthetic protein-assembling entities that function like tRNAs but can read k nucleotides at a time, with k greater than or equal to 5. The realization of this assumption is presented as a challenge to the research community. / Dissertation/Thesis / Masters Thesis Biomedical Engineering 2019

Mutation is a fundamental process that drives evolutionary change; however, most new mutations are deleterious for organismal fitness and can readily propagate within populations under a broad range of conditions. Mutational processes able to counteract deleterious mutation accumulation include: 1) reversion mutation back to wildtype, 2) acquisition of generally beneficial mutations, and 3) compensatory mutations that specifically mitigate the effects of previously-acquired deleterious mutations through epistasis. The potential for any of these mutation types alters our expectations for the impact of deleterious mutation in populations, but since the fitness effects of individual mutations are rarely characterized, the relative importance of beneficial and compensatory epistatic mutations is unknown. In this thesis, I characterized the nuclear mutations that arose in a previous mutation accumulation (MA) experiment using Caenorhabditis elegans nematodes, in which mutations were allowed to accumulate under extreme drift conditions in replicate, independently evolving lines initiated from a low-fitness mitochondrial electron transport chain (ETC) mutant, gas-1. In contrast to the results of typical MA experiments, gas-1 MA lines improved fitness slightly compared to their ancestor. Here, I find that the gas-1 MA lines demonstrate little increase in among-line variance and that the gas-1 MA nuclear mutations are more narrowly functionally defined than wildtype MA nuclear mutations. When combined with evidence for zygotic or post zygotic selection these data suggest that selection--both purifying and positive--can be an extremely powerful force even in conditions of extreme genetic drift. Furthermore, functional characterization of a four-mutation set isolated from one of the gas-1 MA lines on gas-1 and wildtype backgrounds shows fitness improvements on both backgrounds. This beneficial four-mutation set is associated with a decrease in steady-state endogenous ROS on the gas-1 background while exhibiting no effect on wildtype. I also find that steady-state ATP levels associated with the beneficial four-mutation set decreased compared to wildtype suggesting that fermentation may be metabolic strategy to cope with increase oxidative stress. These findings suggest that we can detect and characterize specific genetic changes that lead to a partial recovery of fitness and phenotype in a low-fitness ETC-deficient mutant strain of C. elegans. I extended my thesis to include analyses of fitness and phenotype of 24 replicate lineages of the gas-1 ETC mutant evolved in large population (n = 1000) sizes for 60 generations--conditions optimal for selection and fitness recovery (RC). I find that two distinct gas-1 RC fitness groups emerged: one group with significantly higher average fitness than the ancestor and containing two lines that exceeded wildtype fitness levels, and another group with more modest and non-significant fitness gains. Interestingly, many lines in the first group were observed to generate appreciable numbers of males during experimental evolution--consistent with evolution of outcrossing either accompanying or driving rapid fitness recovery. Bioinformatic functional analyses of the nuclear mutations that arose in the gas-1 RC lines show the availability of potentially more paths to fitness recovery for large populations than small ones. Combined, these data allow us to identify patterns in selection and drift in gas-1 recovery under MA and RC (recovery) conditions. My research advances our understanding of the genetic bases of adaptive evolution under extremely unfavorable population genetic conditions and how mitochondrial dysfunction affects evolutionary dynamics.

Verschiedene Forschungsergebnisse der letzten zehn Jahre ergaben, dass die weitaus häufigeren, nicht-syndromalen Schwerhörigkeiten durch Mutation eines Gens (GJB2-Gen) entstehen, welches im Cortischen Organ des Innenohrs exprimiert wird. Das GJB2-Gen (Connexin-26-Gen), dessen Veränderung etwa 50 % der Fälle von autosomal rezessiver Schwerhörigkeit ausmacht, liegt im Chromosomenbereich 13q11–12. Aktuell identifiziert sind mehr als 70 weitere Loki, die in Verbindung mit nicht-syndromalen Formen von Schwerhörigkeit stehen. Die Prävalenz von NSHL beträgt nach neusten Studien ca. 1,33 pro 1000 Neugeborenen. In Würzburg wurden bis zum Jahr 2011 auf der Neugeborenenstation der Frauenklinik der Universitätsklinik in einem bewährten zweistufigen Neugeborenen-Hörscreening ca. 12853 Babys untersucht. Ziel des Neugeborenen-Hörscreenings ist eine frühestmögliche Erkennung von Schwerhörigkeit bei Neugeborenen, damit durch die Behandlung eine ungehinderte Sprachentwicklung gewährleistet werden kann. In dieser Arbeit wurde der Zusammenhang zwischen der Mutation im Connexin-26-Gen und dem Grad, dem Verlauf und der Konfiguration der Hörminderung untersucht. Hierfür wurden 59 Patienten im Alter von 1 bis 15 Jahren mit beidseitigen, nicht-syndromalen Hörstörungen der Schallempfindung verschiedenen Grades rekrutiert. Mithilfe der molekulargenetischen Befunde konnten Veränderungen im Connexin-26-Gen diagnostiziert werden. Anschließend wurde versucht, unter Zuhilfenahme aller vorhandenen Befunde der individuellen Audiogramm- und BERA- oder ASSR-Befunde eine Genotyp-Phänotyp-Korrelation abzuleiten. / Different research studies in the last ten years resulted, that the more frequently and non-syndromal hearing loss result of mutation in the gene (GJB2-Gen), witch express in cortical organ the inner ear. The GJB2-Gen (Connexin-26-Gen) plays a decisive role in the 50% the autosomal recessive cases of hearing loss and it is localized on chromosome 13q11–12. Up to now there are more than 70 Loci identified, witch also connected with the non-syndromal hearing loss. The prevalence of NSHL amounts according to the newest studies about 1,33 per 1000 newborn. Until 2011, about 12853 babies have been examined at the neonatal ward of the gynecological university hospital in Würzburg in a proven two-stage newborn hearing screening. Goal of the screening was, to detect deafness in neonates as early as possible. Consequently, unhindered language development throughout the treatment can be assured . In this thesis, the relation between mutation in connexin-26 gene and the degree, course and configuration of hearing loss was investigated. For this purpose, 59 patients aged 1 to 15 years with bilateral, non-syndromic hearing loss of a variety acoustic perception form, were gathered. Using molecular genetic examenation methods, changes of the connexin-26 gene were diagnosed. Subsequently, by analyzing all available findings of individual audiogram and BERA or ASSR, a genotype-phenotype correlation was established.

Translation initiation factor IF1 is a small, essential and ubiquitous protein factor encoded by a single infA gene in bacteria. Although several important functions have been attributed to IF1, the precise reason for its indispensability is yet to be defined. It is known that IF1 binds to the ribosomal A-site during initiation, where it primarily contacts ribosomal RNA (rRNA) and induces large scale conformational changes in the small ribosomal subunit. To shed more light on the function of IF1 and its interaction with the ribosome, we have employed a genetic approach to elucidate structure-function interactions between IF1 and rRNA. A selection has been used to isolate second site suppressor mutations in rRNA that restore the growth of a cold sensitive mutant IF1 with an arginine to leucine substitution in position 69 (R69L).  This yielded two classes of suppressors – one class that mapped to the processing stem of 23S rRNA – a transient structure important for proper maturation of 23S rRNA; and the other class to the functional sequence of 16S rRNA. Suppressor mutations in the processing stem of 23S rRNA were shown to disrupt efficient processing of 23S rRNA. In addition, we report that at least one of the manifestations of cold sensitivity associated with the mutant IF1 is at the level of ribosomal subunit association. These results led to a model whereby the cold sensitive R69L mutant IF1 results in aberrant ribosomal subunit association properties, while the 23S processing stem mutations indirectly suppress this effect by decreasing the pool of mature 50S subunits available for association.  Spontaneous suppressor mutations in 16S rRNA were diverse in position and phenotypic properties, but all mutations affected ribosomal subunit association, in most cases by directly decreasing the affinity of the 30S for 50S subunits. Site directed mutagenesis of select positions in 16S rRNA yielded additional suppressor mutations that were localized to the mRNA and streptomycin binding sites on the small ribosomal subunit. We suggest that the 16S rRNA suppressors occur in positions that affect the conformational dynamics brought about by IF1. Taken together, this work indicates that the major function of IF1 is the modulation of ribosomal subunit association brought about through conformational changes of the 30S subunit. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 3: Manuscript.</p>

Long-lived individuals of the opportunistic fungal pathogen Armillaria gallica arise in single mating events, and then grow vegetatively to occupy large territories including multiple woody substrates. In effect, this leaves a spatial record of mutation, the detection of which would allow new inferences about how fungal individuals grow and infect their hosts. In this thesis, I first identified a large individual of A. gallica in eastern Ontario. I then searched for genetic variation within this individual by focusing on the tandemly repeated rRNA gene cluster and four microsatellite markers that are variable in the A. gallica population. I discovered a loss of heterozygosity (LOH) in the rRNA gene-cluster region, forming two genotypes that show significant spatial clustering in a Mantel test. My M.Sc. thesis research serves as a baseline for a genome-wide study of the mutational dynamic within the vegetative growth phase of this large and old Armillaria individual.

Long-lived individuals of the opportunistic fungal pathogen Armillaria gallica arise in single mating events, and then grow vegetatively to occupy large territories including multiple woody substrates. In effect, this leaves a spatial record of mutation, the detection of which would allow new inferences about how fungal individuals grow and infect their hosts. In this thesis, I first identified a large individual of A. gallica in eastern Ontario. I then searched for genetic variation within this individual by focusing on the tandemly repeated rRNA gene cluster and four microsatellite markers that are variable in the A. gallica population. I discovered a loss of heterozygosity (LOH) in the rRNA gene-cluster region, forming two genotypes that show significant spatial clustering in a Mantel test. My M.Sc. thesis research serves as a baseline for a genome-wide study of the mutational dynamic within the vegetative growth phase of this large and old Armillaria individual.