FPGA-based Acceleration of Detecting Statistical Epistasis in GWAS

Wienbrandt, Lars; Kässens, Jan Christian; González-Domínguez, Jorge; Schmidt, Bertil; Ellinghaus, David; Schimmler, Manfred

doi:10.1016/j.procs.2014.05.020

Cited by 29 publications

(15 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hadoop/Spark is designed to operate on commodity machines, and it should therefore be kept in mind that alternative computing frameworks such as High Performance Computing solutions, or those based on dedicated hardware such as GPUs [72,71,25,18] and FPGAs [68,20], still retain the edge in terms of raw computational capacity. Rather, the main benefits of Hadoop/Spark lie in its robustness to node failure, and its ability to provide an abstraction of the distributed infrastructure.…”

Section: Discussion and Research Perspectivesmentioning

confidence: 99%

“…Few solutions have been designed to address this challenge, that mostly rely on dedicated hardware devices such as Graphical Processing Units (GPUs) [72,71,25,18], or Field-Programmable Gate Array (FPGAs) [68,20]. While these solutions greatly speed up computation times, their use is in practice hindered by the need to acquire specialised and expensive hardware, whose programming is based on low-level and difficult to debug programming languages.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DiGeST: Distributed Computing for Scalable Gene and Variant Ranking with Hadoop/Spark

Helaers

Lenaerts

et al. 2017

Preprint

View full text Add to dashboard Cite

Background:The advent of next-generation sequencing technologies has opened new avenues for clinical genomics research. In particular, as sequencing costs continue to decrease, an ever-growing number of clinical genomics institutes now rely on DNA sequencing studies at varying scales -genome, exome, mendeliome -for uncovering disease-associated variants or genes, in both rare and non-rare diseases.A common methodology for identifying such variants or genes is to rely on genetic association studies (GAS), that test whether allele or genotype frequencies differ between two groups of individuals, usually diseased subjects and healthy controls. Current bioinformatics tools for performing GAS are designed to run on standalone machines, and do not scale well with the increasing size of study designs and the search for multi-locus genetic associations. More efficient distributed and scalable data analysis solutions are needed to address this challenge. Results: We developed a Big Data solution stack for distributing computations in genetic association studies, that address both single and multi-locus associations. The proposed stack, called DiGeST (Distributed Gene/variant Scoring Tool) is divided in two main components: a Hadoop/Spark high-performance computing back-end for efficient data storage and distributed computing, and a Web front-end providing users with a rich set of options to filter, compare and explore exome data from different sample populations. Using exome data

show abstract

Section: Discussion and Research Perspectivesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

DiGeST: Distributed Computing for Scalable Gene and Variant Ranking with Hadoop/Spark

Helaers

Lenaerts

et al. 2017

Preprint

View full text Add to dashboard Cite

show abstract

“…However, these works are limited to pairwise interactions. The target architectures vary from one to another, including GPUs (González-Domínguez et al, 2014; Goudey et al, 2013; Hemani et al, 2011; Hu et al, 2010; Piriyapongsa et al, 2012; Yung et al, 2011), field-programmable gate arrays (FPGAs) (Wienbrandt et al, 2014), multi-node clusters (Kässens et al, 2014; Ma et al, 2008; Schüpbach et al, 2010), or cloud-based architectures (Yoshizoe et al, 2018; Zhou et al, 2016).…”

Section: Related Workmentioning

confidence: 99%

Fast search of third-order epistatic interactions on CPU and GPU clusters

Ponte-Fernández

González-Domínguez

Martín

2019

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

Genome-Wide Association Studies (GWASs), analyses that try to find a link between a given phenotype (such as a disease) and genetic markers, have been growing in popularity in the recent years. Relations between phenotypes and genotypes are not easy to identify, as most of the phenotypes are a product of the interaction between multiple genes, a phenomenon known as epistasis. Many authors have resorted to different approaches and hardware architectures in order to mitigate the exponential time complexity of the problem. However, these studies make some compromises in order to keep a reasonable execution time, such as limiting the number of genetic markers involved in the interaction, or discarding some of these markers in an initial filtering stage. This work presents MPI3SNP, a tool that implements a three-way exhaustive search for cluster architectures with the aim of mitigating the exponential growth of the run-time. Modern cluster solutions usually incorporate GPUs. Thus, MPI3SNP includes implementations for both multi-CPU and multi-GPU clusters. To contextualize the performance achieved, MPI3SNP is able to analyze an input of 6300 genetic markers and 3200 samples in less than 6 min using 768 CPU cores or 4 min using 8 NVIDIA K80 GPUs. The source code is available at https://github.com/chponte/mpi3snp .

show abstract

“…The FPGA pipeline for contingency table generation is based on our previous work for pairwise [12,25] and third-order interactions [26]. Thus, we omit details here and only remark the differences.…”

Section: Task Distributionmentioning

confidence: 99%

1,000x Faster than PLINK: Genome-Wide Epistasis Detection with Logistic Regression Using Combined FPGA and GPU Accelerators

Wienbrandt

Kässens

Hübenthal

et al. 2018

EasyChair Preprints

Self Cite

View full text Add to dashboard Cite

Abstract. Logistic regression as implemented in PLINK is a powerful and commonly used framework for assessing gene-gene (GxG) interactions. However, fitting regression models for each pair of markers in a genome-wide dataset is a computationally intensive task. Performing billions of tests with PLINK takes days if not weeks, for which reason prefiltering techniques and fast epistasis screenings are applied to reduce the computational burden. Here, we demonstrate that employing a combination of a Xilinx UltraScale KU115 FPGA with an Nvidia Tesla P100 GPU leads to runtimes of only minutes for logistic regression GxG tests on a genome-wide scale. In particular, a dataset of 53,000 samples genotyped at 130,000 SNPs was analyzed in 8 minutes, resulting in a speedup of more than 1,000 when compared to PLINK v1.9 using 32 threads on a server-grade computing platform. Furthermore, on-the-fly calculation of test statistics, p-values and LD-scores in double-precision make commonly used prefiltering strategies obsolete.

show abstract

FPGA-based Acceleration of Detecting Statistical Epistasis in GWAS

Cited by 29 publications

References 23 publications

DiGeST: Distributed Computing for Scalable Gene and Variant Ranking with Hadoop/Spark

DiGeST: Distributed Computing for Scalable Gene and Variant Ranking with Hadoop/Spark

Fast search of third-order epistatic interactions on CPU and GPU clusters

1,000x Faster than PLINK: Genome-Wide Epistasis Detection with Logistic Regression Using Combined FPGA and GPU Accelerators

Contact Info

Product

Resources

About