Error Correction And Clustering Algorithms For Next Generation Sequencing

Author : Raunaq Malhotra
Publisher :
ISBN 13 :
Total Pages : pages
Book Rating : 4.:/5 (959 download)

DOWNLOAD NOW!

Download or read book De Novo Methods for Characterizing Diversity in Populations of Genomes Using Next-generation Sequencing Data written by Raunaq Malhotra and published by . This book was released on 2016 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: Next-generation sequencing (NGS) technologies have enabled fast profiling of diversity in population of closely related genomes over the past two decades in a high throughput fashion. Examples of such applications include reconstructing the haplotypes of viruses replicating within a host, identifying insertional polymorphisms of mobile elements amongst individuals in a species, and characterizing diversity in populations of cancer cell types in an individual. Although a plethora of methods use an existing or assembled reference sequence for characterizing diversity, the usage of de novo methods can be beneficial for studying diversity in a population of closely related genomes. This dissertation aims to develop de novo computational methods for studying diversity in such samples using NGS data. De novo methods rely only on the inherent NGS data collected from a sample and can be applied to study the diversity in organisms where limited prior knowledge is available. In the first part of the dissertation, I discuss a clustering pipeline for clustering NGS data that was obtained from sites of integration of a mobile element in the genome in a panel of individuals. Typically, reads from individuals can be mapped to a reference genome to determine the location of an element-host integration site. However, host regions flanking a mobile element can be modified by the presence of the element and reference genomes are not available for all species. Clustering methods provide an alternative for analyzing such integration site datasets. Here, each cluster ideally represents reads from a single integration site. The main contributions in this chapter are (i) A pipeline for clustering NGS reads from integration site junctions using UCLUST clustering algorithm that empirically determines the optimal clustering threshold; (ii) The optimal threshold is determined based on internal clustering measure, $I-index$, which assesses clusters for small intra-cluster diameters and large inter-cluster distance. We evaluate and test our proposed pipeline to determine the optimal clustering parameters and show improved results on simulated and real datasets. The second and major portion of the dissertation focuses on de novo reconstruction of viral haplotypes in a viral population using paired-end NGS data. We have proposed MLEHaplo, a maximum likelihood de novo assembly algorithm for reconstructing viral haplotypes. Using the pairing information of reads in our proposed Viral Path Reconstruction Algorithm (ViPRA), we generate a small subset of paths from a De Bruijn graph of reads that serve as candidate paths for true viral haplotypes. Our proposed method MLEHaplo then generates a maximum likelihood estimate of the viral population using the paths reconstructed by ViPRA. We evaluate and compare MLEHaplo on simulated datasets at different sequence coverage, and on real datasets. MLEHaplo reconstructs full length viral haplotypes in most of the small genome simulated viral populations. While reference based methods either under-estimate or over-estimate the viral haplotypes, MLEHaplo limits the over-estimation of the size of true viral haplotypes, and reconstructs the full phylogeny of the input viral population. As NGS data is fraught with sequencing errors which significantly impact de novo reconstruction of viral haplotypes, in Chapter 4 we proposed a frame-based representation of k-mers for detecting sequencing errors and rare variants for such data. Frames are sets of non-orthogonal basis functions, traditionally used in signal processing for noise removal. We define a frame for genomes and sequenced reads to consist of discrete spatial signals of every k-mer of a given size. We show that each k-mer in the sequenced data can be projected onto multiple frames and these projections are maximized for spatial signals corresponding to the k-mer's substrings. Our proposed classifier, MultiRes, is trained on the projections of k-mers as features used for marking k-mers as erroneous or true variations in the genome. We evaluate MultiRes on simulated and real viral population datasets and compare it to other error correction methods known in the literature. MultiRes has 4 to 500 times less false positive k-mer predictions compared to other methods, essential for accurate estimation of viral population diversity and their de novo assembly. It has high recall of the true k-mers, comparable to other error correction methods. MultiRes also has greater than 95% recall for detecting single nucleotide polymorphisms (SNPs), has low false positive SNP predictions, while detecting higher number of rare variants compared to other variant calling methods for viral populations.