Fig 1 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Typical variant calling bioinformatics pipeline composed of steps following NGS sequencing leading to the visualization of data is presented in the middle panel. The variant calling bioinformatics pipeline is contained within the data pre-processing and data analysis stages of a much larger bioinformatics-based research study as illustrated in the upper panel. Data file formats at each step are presented in the lower panel (Al Kawam et al., 2017; Lightbody et al., 2019). *Platform-specific raw sequence output either .BAM or .FASTQ or .HDF5 (NCBI, 2019).
Source publication
The significant decline in the cost of genome sequencing has dramatically changed the typical bioinformatics pipeline for analysing sequencing data. Where traditionally, the computational challenge of sequencing is now secondary to genomic data analysis. Short read alignment (SRA) is a ubiquitous process within every modern bioinformatics pipeline...
Contexts in source publication
Context 1
... bioinformatics based research study typically consists of study design, sample collection, library preparation to eventual NGS sequencing and data analysis (Fig. 1, upper panel) ( Lightbody et al., 2019). Within which, a typical bioinformatics pipeline represents data preprocessing and data analysis workflows actioned to yield useable insights from sequenced samples. Such workflows are typically dependent upon the end application, such as variant calling (Fig. 1, middle panel), and thus, overall ...
Context 2
... to eventual NGS sequencing and data analysis (Fig. 1, upper panel) ( Lightbody et al., 2019). Within which, a typical bioinformatics pipeline represents data preprocessing and data analysis workflows actioned to yield useable insights from sequenced samples. Such workflows are typically dependent upon the end application, such as variant calling (Fig. 1, middle panel), and thus, overall study design. However, they share some common steps such as quality control, alignment, pre-and post-alignment filtering and visualization. Each step has its own unique set of barriers and facilitating factors, which have an ultimate bearing on the quality of data output for analysis ( Lightbody et ...
Context 3
... there exists an optimum network size depending on the alignment algorithm used; e.g. BWA scales linearly, whereas HISAT2 (Kim et al., 2019) shows a decay in execution speed with higher than 4 x 4 network sizes. This is further illustrated by Das and Ghosal (2018) who suggest that the NoC network topology, particularly those relying exclusively on mesh topologies, result in higher latency (slower performance) at higher network dimensions. ...
Citations
... Recently, there have been multiple efforts to accelerate widely-used genome analysis kernels exploiting novel hardware solutions [91][92][93][94][95][96]. Tony et al. [97] provide a comprehensive review of state-of-the-art hardware acceleration techniques for genomics. ...
Arm usage has substantially grown in the High-Performance Computing (HPC) community. Japanese supercomputer Fugaku, powered by Arm-based A64FX processors, held the top position on the Top500 list between June 2020 and June 2022, currently sitting in the fourth position. The recently released 7th generation of Amazon EC2 instances for compute-intensive workloads (C7 g) is also powered by Arm Graviton3 processors. Projects like European Mont-Blanc and U.S. DOE/NNSA Astra are further examples of Arm irruption in HPC. In parallel, over the last decade, the rapid improvement of genomic sequencing technologies and the exponential growth of sequencing data has placed a significant bottleneck on the computational side. While most genomics applications have been thoroughly tested and optimized for x86 systems, just a few are prepared to perform efficiently on Arm machines. Moreover, these applications do not exploit the newly introduced Scalable Vector Extensions (SVE).
This paper presents GenArchBench, the first genome analysis benchmark suite targeting Arm architectures. We have selected computationally demanding kernels from the most widely used tools in genome data analysis and ported them to Arm-based A64FX and Graviton3 processors. Overall, the GenArch benchmark suite comprises 13 multi-core kernels from critical stages of widely-used genome analysis pipelines, including base-calling, read mapping, variant calling, and genome assembly. Our benchmark suite includes different input data sets per kernel (small and large), each with a corresponding regression test to verify the correctness of each execution automatically. Moreover, the porting features the usage of the novel Arm SVE instructions, algorithmic and code optimizations, and the exploitation of Arm-optimized libraries. We present the optimizations implemented in each kernel and a detailed performance evaluation and comparison of their performance on four different HPC machines (i.e., A64FX, Graviton3, Intel Xeon Skylake Platinum, and AMD EPYC Rome). Overall, the experimental evaluation shows that Graviton3 outperforms other machines on average. Moreover, we observed that the performance of the A64FX is significantly constrained by its small memory hierarchy and latencies. Additionally, as proof of concept, we study the performance of a production-ready tool that exploits two of the ported and optimized genomic kernels.
... This situation has resulted in an increasing computational bottleneck in the genome analysis pipelines [8]. To face this challenge, an increasing number of algorithms [4] and cutting-edge hardware accelerators [9] have been developed to accelerate the genome analysis pipelines. ...
The RISC-V ISA has gained significant momentum in High-Performance Computing (HPC) research and market due to its open-source nature, fostering collaborative research and innovation. The ever-growing RISC-V-based hardware/software ecosystem has made it an attractive option for HPC application development and production. Within the field of biomedical research, genome data analysis has emerged as a crucial step towards personalized medicine, demanding substantial computational resources and more efficient tools.
This paper presents a benchmark suite of genome analysis kernels ported to RISC-V and their evaluation on modern RISC-V systems.
Our work evaluates the RISC-V toolchain's maturity and the software/hardware ecosystem's readiness for its adoption for genome data analysis. This study aims to provide valuable guidance for researchers and practitioners interested in adopting RISC-V for genome analysis, and provides feedback to the RISC-V community on the challenges that need to be addressed for RISC-V to become an efficient HPC platform.
... Recent powerful graphics processing units (GPUs) can expedite fast deep learning inference by enabling massive parallel computation [3]. However, it is challenging to maintain large in-house GPU facilities in a small hospital or biomedical research laboratory due to the high initial equipment costs and significant power consumption [4]. ...
Convolutional neural networks (CNNs) have enabled effective object detection tasks in bioimages. Unfortunately, implementing such an object detection model can be computationally intensive, especially on resource-limited hardware in a laboratory or hospital setting. This study aims to develop a framework called BioEdge that can accelerate object detection using Scaled-YOLOv4 and YOLOv7 by leveraging edge computing for bioimage analysis. BioEdge employs a distributed inference technique with Scaled-YOLOv4 and YOLOv7 to harness the computational resources of both a local computer and an edge server, enabling rapid detection of COVID-19 abnormalities in chest radiographs. By implementing distributed inference techniques, BioEdge addresses privacy concerns that can arise when transmitting biomedical data to an edge server. Additionally, it incorporates a computationally lightweight autoencoder at the split point to reduce data transmission overhead. For evaluation, this study utilizes the COVID-19 dataset provided by the Society for Imaging Informatics in Medicine (SIIM). BioEdge is shown to improve the inference latency of Scaled-YOLOv4 and YOLOv7 by up to 6.28 times with negligible accuracy loss compared to local computer execution in our evaluation setting.
... In addition to the potential of GPU-enabled algorithms, an increasing number of FPGA devices are available for precision medicine variant detection (60). FPGAs are integrated circuits designed to be configured for specific software applications. ...
Precision medicine programs to identify clinically relevant genetic variation have been revolutionized by access to increasingly affordable high-throughput sequencing technologies. A decade of continual drops in per-base sequencing costs means it is now feasible to sequence an individual patient genome and interrogate all classes of genetic variation for < $1,000 USD. However, while advances in these technologies have greatly simplified the ability to obtain patient sequence information, the timely analysis and interpretation of variant information remains a challenge for the rollout of large-scale precision medicine programs. This review will examine the challenges and potential solutions that exist in identifying predictive genetic biomarkers and pharmacogenetic variants in a patient and discuss the larger bioinformatic challenges likely to emerge in the future. It will examine how both software and hardware development are aiming to overcome issues in short read mapping, variant detection and variant interpretation. It will discuss the current state of the art for genetic disease and the remaining challenges to overcome for complex disease. Success across all types of disease will require novel statistical models and software in order to ensure precision medicine programs realize their full potential now and into the future.
... The focus of this dissertation has been on scalable distributed and parallel computing methods on CPUs. In addition, GPGPU [54,75,218] and FPGA [55,76,219] computing offer complementary approaches for increasing the performance with fine-grained data parallelism nested with coarse-grained data-parallel computing on CPUs in complex bioinformatics pipelines and algorithms, which is a promising area of our future research. ...
High-throughput sequencing (HTS) technologies have enabled rapid DNA sequencing of whole-genomes collected from various organisms and environments, including human tissues, plants, soil, water, and air. As a result, sequencing data volumes have grown by several orders of magnitude, and the number of assembled whole-genomes is increasing rapidly as well. This whole-genome sequencing (WGS) data has revealed the genetic variation in humans and other species, and advanced various fields from human and microbial genomics to drug design and personalized medicine. The amount of sequencing data has almost doubled every six months, creating new possibilities but also big data challenges in genomics. Diverse methods used in modern computational biology require a vast amount of computational power, and advances in HTS technology are even widening the gap between the analysis input data and the analysis outcome.
Currently, many of the existing genomic analysis tools, algorithms, and pipelines are not fully exploiting the power of distributed and high-performance computing, which in turn limits the analysis throughput and restrains the deployment of the applications to clinical practice in the long run. Thus, the relevance of harnessing distributed and cloud computing in bioinformatics is more significant than ever before. Besides, efficient data compression and storage methods for genomic data processing and retrieval integrated with conventional bioinformatics tools are essential. These vast datasets have to be stored and structured in formats that can be managed, processed, searched, and analyzed efficiently in distributed systems.
Genomic data contain repetitive sequences, which is one key property in developing efficient compression algorithms to alleviate the data storage burden. Moreover, indexing compressed sequences appropriately for bioinformatics tools, such as read aligners, offers direct sequence search and alignment capabilities with compressed indexes. Relative Lempel-Ziv (RLZ) has been found to be an efficient compression method for repetitive genomes that complies with the data-parallel computing approach. RLZ has recently been used to build hybrid-indexes compatible with read aligners, and we focus on extending it with distributed computing. Data structures found in genomic data formats have properties suitable for parallelizing routine bioinformatics methods, e.g., sequence matching, read alignment, genome assembly, genotype imputation, and variant calling. Compressed indexing fused with the routine bioinformatics methods and data-parallel computing seems a promising approach to building population-scale genome analysis pipelines. Various data decomposition and transformation strategies are studied for optimizing data-parallel computing performance when such routine bioinformatics methods are executed in a complex pipeline. These novel distributed methods are studied in this dissertation and demonstrated in a generalized scalable bioinformatics analysis pipeline design.
The dissertation starts from the main concepts of genomics and DNA sequencing technologies and builds routine bioinformatics methods on the principles of distributed and parallel computing. This dissertation advances towards designing fully distributed and scalable bioinformatics pipelines focusing on population genomic problems where the input data sets are vast and the analysis results are hard to achieve with conventional computing. Finally, the methods studied are applied in scalable population genomics applications using real WGS data and experimented with in a high performance computing cluster. The experiments include mining virus sequences from human metagenomes, imputing genotypes from large-scale human populations, sequence alignment with compressed pan-genomic indexes, and assembling reference genomes for pan-genomic variant calling.
Sequence alignment pipelines for human genomes are an emerging workload that will dominate in the precision medicine field. BWA-MEM2 is a tool widely used in the scientific community to perform read mapping studies. In this paper, we port BWA-MEM2 to the AArch64 architecture using the ARMv8-A specification, and we compare the resulting version against an Intel Skylake system both in performance and in energy-to-solution. The porting effort entails numerous code modifications, since BWA-MEM2 implements certain kernels using x86_64 specific intrinsics, e.g., AVX-512. To adapt this code we use the recently introduced Arm's Scalable Vector Extensions (SVE). More specifically, we use Fujitsu's A64FX processor, the first to implement SVE. The A64FX powers the Fugaku Supercomputer that led the Top500 ranking from June 2020 to November 2021. After porting BWA-MEM2 we define and implement a number of optimizations to improve performance in the A64FX target architecture. We show that while the A64FX performance is lower than that of the Skylake system, A64FX delivers 11.6% better energy-to-solution on average. All the code used for this article is available at
https://gitlab.bsc.es/rlangari/bwa-a64fx
.
The study of multiple “omes,” such as the genome, transcriptome, proteome, and metabolome has become widespread in biomedical research. High‐throughput techniques enable the rapid generation of high‐dimensional multiomics data. This multiomics approach provides a more complete perspective to study biological systems compared with traditional methods. However, the quantitative analysis and integration of distinct types of high‐dimensional omics data remain a challenge. Here, we provide an up‐to‐date and comprehensive review of the methods used for omics data quantification and integration. We first review the quantitative analysis of not only bulk but also single‐cell transcriptomics data, as well as proteomics data. Current methods for reducing batch effects and integrating heterogeneous high‐dimensional data are then introduced. Network analysis on large‐scale biomedical data can capture the global properties of drugs, targets, and disease relationships, thus enabling a better understanding of biological systems. Current trends in the applications and methods used to extend quantitative omics data analysis to biological networks are also discussed. This article is categorized under: Data Science > Artificial Intelligence/Machine Learning This review provides an up‐to‐date and comprehensive overview of the methods for omics data quantification and integration, as well as their applications.
In view of the problems of long warning time and poor protection effect of traditional gateway boundary security protection platform, a gateway boundary security protection platform based on Internet of things and cloud computing is designed. Net FPGA chip is used for the verification and development of network communication equipment, connecting the ATA serial port connection line port of multiple boards. Combined with the register host computer, the read-write operation of the registers inside each module in the hardware is completed through PCI bus, and the hardware design of the gateway boundary security protection platform is completed. Establish the gateway border security protection module and complete the software design of the gateway border security protection platform. Based on the Internet of things and cloud computing technology, match the network security link, so as to realize the security protection of the network boundary. The experimental results show that the security protection effect of the platform constructed in this paper is better, and can effectively shorten the security early warning time.
