Assembling Large Genomes with Single-Molecule Sequencing and Locality Sensitive Hashing

Konstantin Berlin, Sergey Koren, Chen-Shan Chin, James Drake, Jane M Landolin, Adam M Phillippy
doi: http://dx.doi.org/10.1101/008003

We report reference-grade de novo assemblies of four model organisms and the human genome from single-molecule, real-time (SMRT) sequencing. Long-read SMRT sequencing is routinely used to finish microbial genomes, but the available assembly methods have not scaled well to larger genomes. Here we introduce the MinHash Alignment Process (MHAP) for efficient overlapping of noisy, long reads using probabilistic, locality-sensitive hashing. Together with Celera Assembler, MHAP was used to reconstruct the genomes of Escherichia coli, Saccharomyces cerevisiae, Arabidopsis thaliana, Drosophila melanogaster, and human from high-coverage SMRT sequencing. The resulting assemblies include fully resolved chromosome arms and close persistent gaps in these important reference genomes, including heterochromatic and telomeric transition sequences. For D. melanogaster, MHAP achieved a 600-fold speedup relative to prior methods and a cloud computing cost of a few hundred dollars. These results demonstrate that single-molecule sequencing alone can produce near-complete eukaryotic genomes at modest cost.

Efficient Bayesian mixed model analysis increases association power in large cohorts

Po-Ru Loh, George Tucker, Brendan K Bulik-Sullivan, Bjarni J Vilhjalmsson, Hilary K Finucane, Daniel I Chasman, Paul M Ridker, Benjamin M Neale, Bonnie Berger, Nick Patterson, Alkes L Price
doi: http://dx.doi.org/10.1101/007799

Linear mixed models are a powerful statistical tool for identifying genetic associations and avoiding confounding. However, existing methods are computationally intractable in large cohorts, and may not optimize power. All existing methods require time cost O(MN^2) (where N = #samples and M = #SNPs) and implicitly assume an infinitesimal genetic architecture in which effect sizes are normally distributed, which can limit power. Here, we present a far more efficient mixed model association method, BOLT-LMM, which requires only a small number of O(MN) iterations and increases power by modeling more realistic, non-infinitesimal genetic architectures via a Bayesian mixture prior on marker effect sizes. We applied BOLT-LMM to nine quantitative traits in 23,294 samples from the Women’s Genome Health Study (WGHS) and observed significant increases in power, consistent with simulations. Theory and simulations show that the boost in power increases with cohort size, making BOLT-LMM appealing for GWAS in large cohorts.

Genome-wide predictability of restriction sites across the eukaryotic tree of life

Santiago Herrera, Paula H. Reyes-Herrera, Timothy M. Shank
doi: http://dx.doi.org/10.1101/007781

High-throughput sequencing of reduced representation libraries obtained through digestion with restriction enzymes–generally known as restriction-site associated DNA sequencing (RAD-seq)–is now one most commonly used strategies to generate single nucleotide polymorphism data in eukaryotes. The choice of restriction enzyme is critical for the design of any RAD-seq study as it determines the number of genetic markers that can be obtained for a given species, and ultimately the success of a project. In this study we tested the hypothesis that genome composition, in terms of GC content, mono-, di- and trinucleotide compositions, can be used to predict the number of restriction sites for a given combination of restriction enzyme and genome. We performed systematic in silico genome-wide surveys of restriction sites across the eukaryotic tree of live and compared them with expectations generated from stochastic models based on genome compositions using the newly developed software pipeline PredRAD (https://github.com/phrh/PredRAD). Our analyses reveal that in most cases the trinucleotide genome composition model is the best predictor, and the GC content and mononucleotide models are the worst predictors of the expected number of restriction sites in a eukaryotic genome. However, we argue that the predictability of restriction site frequencies in eukaryotic genomes needs to be treated in a case-specific basis, because the phylogenetic position of the taxon of interest and the specific recognition sequence of the selected restriction enzyme are the most determinant factors. The results from this study, and the software developed, will help guide the design of any study using RAD sequencing and related methods.

Benchmarking undedicated cloud computing providers for analysis of genomic datasets.

Seyhan Yazar, George EC Gooden, David A Mackey, Alex Hewitt
doi: http://dx.doi.org/10.1101/007724

A major bottleneck in biological discovery is now emerging at the computational level. Cloud computing offers a dynamic means whereby small and medium-sized laboratories can rapidly adjust their computational capacity. We benchmarked two established cloud computing services, Amazon Web Services Elastic MapReduce (EMR) on Amazon EC2 instances and Google Compute Engine (GCE), using publicly available genomic datasets (E.coli CC102 strain and a Han Chinese male genome) and a standard bioinformatic pipeline on a Hadoop-based platform. Wall-clock time for complete assembly differed by 52.9% (95%CI: 27.5-78.2) for E.coli and 53.5% (95%CI: 34.4-72.6) for human genome, with GCE being more efficient than EMR. The cost of running this experiment on EMR and GCE differed significantly, with the costs on EMR being 257.3% (95%CI: 211.5-303.1) and 173.9% (95%CI: 134.6-213.1) more expensive for E.coli and human assemblies respectively. Thus, GCE was found to outperform EMR both in terms of cost and wall-clock time. Our findings confirm that cloud computing is an efficient and potentially cost-effective alternative for analysis of large genomic datasets. In addition to releasing our cost-effectiveness comparison, we present available ready-to-use scripts for establishing Hadoop instances with Ganglia monitoring on EC2 or GCE.

Calling genotypes from public RNA-sequencing data enables identification of genetic variants that affect gene-expression levels

Patrick Deelen, Daria Zhernakova, Mark de Haan, Marijke van der Sijde, Marc Jan Bonder, Juha Karjalainen, K. Joeri van der Velde, Kristin M. Abbott, Jingyuan Fu, Cisca Wijmenga, Richard J. Sinke, Morris A. Swertz, Lude Franke
doi: http://dx.doi.org/10.1101/007633

Given increasing numbers of RNA-seq samples in the public domain, we studied to what extent expression quantitative trait loci (eQTLs) and allele-specific expression (ASE) can be identified in public RNA-seq data while also deriving the genotypes from the RNA-seq reads. 4,978 human RNA-seq runs, representing many different tissues and cell-types, passed quality control. Even though this data originated from many different laboratories, samples reflecting the same cell-type clustered together, suggesting that technical biases due to different sequencing protocols were limited. We derived genotypes from the RNA-seq reads and imputed non-coding variants. In a joint analysis on 1,262 samples combined, we identified cis-eQTLs effects for 8,034 unique genes. Additionally, we observed strong ASE effects for 34 rare pathogenic variants, corroborating previously observed effects on the corresponding protein levels. Given the exponential growth of the number of publicly available RNA-seq samples, we expect this approach will become relevant for studying tissue-specific effects of rare pathogenic genetic variants.