Residual Neural Network for Predicting Super-Enhancers on Genome Scale

Sabba, Sara; Smara, Meroua; Benhacine, Mehdi; Hameurlaine, Amina

doi:10.1007/978-3-030-96311-8_4

Cited by 4 publications

(5 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For , we demonstrate its applicability on basecalling only, while there are other genome sequencing tasks where deep learning models with skip connections are actively being developed, such as predicting the effect of genetic variations [73, 81], detecting replication dynamics [82], and predicting super-enhancers [83]. In Supplementary S2.1, we show the effect of manual skip removal, where we manually remove all the skip connections at once.…”

Section: Discussionmentioning

confidence: 99%

A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

Singh

Alser

Khodamoradi³

et al. 2022

Preprint

View full text Add to dashboard Cite

Nanopore sequencing is a widely-used high-throughput genome sequencing technology that can sequence long fragments of a genome. Nanopore sequencing generates noisy electrical signals that need to be converted into a standard string of DNA nucleotide bases (i.e., A, C, G, T) using a computational step called basecalling. The accuracy and speed of basecalling have critical implications for every subsequent step in genome analysis. Currently, basecallers are mainly based on deep learning techniques to provide high sequencing accuracy without considering the compute demands of such tools. We observe that state-of-the-art basecallers (i.e., Guppy, Bonito) are slow, inefficient, and memory-hungry as researchers have adapted deep learning models from other domains without specialization to the basecalling purpose. Our goal is to make basecalling highly efficient and fast by building the first framework for specializing and optimizing machine learning-based basecaller. We introduce RUBICON, a framework to develop hardware-optimized basecallers. RUBICON consists of two novel machine-learning techniques that are specifically designed for basecalling. First, we introduce the quantization-aware basecalling neural architecture search (QABAS) framework to specialize the basecalling neural network architecture for a given hardware acceleration platform while jointly exploring and finding the best bit-width precision for each neural network layer. Second, we develop SkipClip, the first technique to remove all the skip connections present in modern basecallers to greatly reduce resource and storage requirements without any loss in basecalling accuracy. We demonstrate the benefits of QABAS and SkipClip by developing RUBICALL, the first hardware-optimized basecaller that performs fast and accurate basecalling. Our experimental results on state-of-the-art computing systems show that RUBICALL is a fast, accurate and hardware-friendly, mixed-precision basecaller. Compared to a highly-accurate state-of-the-art basecaller, RUBICALL provides a 16.56x speedup without losing accuracy, while also achieving a 6.88x and 2.94x reduction in neural network model size and the number of parameters, respectively. Compared to the fastest state-of-the-art basecaller, RUBICALL provides a 3.19x speedup with 2.97% higher accuracy. We show that QABAS and SkipClip can help researchers develop hardware-optimized basecallers that are superior to expert-designed models.

show abstract

Section: Discussionmentioning

confidence: 99%

A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

Singh

Alser

Khodamoradi³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…For SkipClip, we demonstrate its applicability on basecalling only, while there are other genome sequencing tasks where deep learning models with skip connections are actively being developed, such as predicting the e ect of genetic variations [69,82], detecting replication dynamics [83], and predicting super-enhancers [84]. In Supplementary S1, we show the e ect of manual skip removal, where we manually remove all the skip connections at once.…”

Section: Discussionmentioning

confidence: 96%

A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

Singh¹,

Alser²,

Khodamoradi³

et al. 2022

Preprint

View full text Add to dashboard Cite

Nanopore sequencing is a widely-used high-throughput genome sequencing technology that can sequence long fragments of a genome. Nanopore sequencing generates noisy electrical signals that need to be converted into a standard string of DNA nucleotide bases (i.e., A, C, G, T) using a computational step called basecalling. The accuracy and speed of basecalling have critical implications for every subsequent step in genome analysis. Currently, basecallers are mainly based on deep learning techniques to provide high sequencing accuracy without considering the compute demands of such tools. We observe that state-of-theart basecallers (i.e., Guppy, Bonito) are slow, ine cient, and memory-hungry as researchers have adapted deep learning models from other domains without specialization to the basecalling purpose. Our goal is to make basecalling highly e cient and fast by building the rst framework for specializing and optimizing machine learning-based basecaller. We introduce RUBICON, a framework to develop hardware-optimized basecallers. RUBICON consists of two novel machine-learning techniques that are speci cally designed for basecalling. First, we introduce the quantization-aware basecalling neural architecture search (QABAS) framework to specialize the basecalling neural network architecture for a given hardware acceleration platform while jointly exploring and nding the best bit-width precision for each neural network layer. Second, we develop SkipClip, the rst technique to remove all the skip connections present in modern basecallers to greatly reduce resource and storage requirements without any loss in basecalling accuracy. We demonstrate the bene ts of QABAS and SkipClip by developing RUBICALL, the rst hardware-optimized basecaller that performs fast and accurate basecalling. Our experimental results on state-of-the-art computing systems show that RUBICALL is a fast, accurate and hardware-friendly, mixed-precision basecaller. Compared to a highly-accurate state-of-the-art basecaller, RUBICALL provides 16.56× speedup without losing accuracy, while also achieving a 6.88× and 2.94× reduction in neural network model size and the number of parameters, respectively. Compared to the fastest state-of-the-art basecaller, RUBICALL provides 3.19× speedup and provides 2.97% higher accuracy. We show that QABAS and SkipClip can help researchers develop hardware-optimized basecallers that are superior to expert-designed models.

show abstract

“…Subsequently, aiming to improve the DEEPSEN method, Sabba et al . developed the ResSEN method using a residual neural network (ResNet) to optimize CNN performance and avoid the vanishing gradient problem ( Table 3 ) [ 95 ]. ResNet is a method of learning deeper CNN by residual learning that introduces residual blocks to avoid degradation problems caused by vanishing gradient problems and other problems that occur when layers are deepened in CNNs [ 96 , 97 ].…”

Section: Introductionmentioning

confidence: 99%

“…In order to validate the performance of the ResSEN method, a comparison was made between the ResSEN and DEEPSEN methods using the same data as the DEEPSEN method, with 90% of the data for training and 10% of the data for testing. As a result, overfitting was observed in the DEEPSEN method but not in the ResSEN method [ 95 ]. The accuracy was also better when predicted by the ResSEN method, indicating that changing the algorithm used for ResNet increases the prediction accuracy of the SE [ 95 ].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Analysis of super-enhancer using machine learning and its application to medical biology

Hamamoto

Takasawa

Shinkai

et al. 2023

Briefings in Bioinformatics

View full text Add to dashboard Cite

The analysis of super-enhancers (SEs) has recently attracted attention in elucidating the molecular mechanisms of cancer and other diseases. SEs are genomic structures that strongly induce gene expression and have been reported to contribute to the overexpression of oncogenes. Because the analysis of SEs and integrated analysis with other data are performed using large amounts of genome-wide data, artificial intelligence technology, with machine learning at its core, has recently begun to be utilized. In promoting precision medicine, it is important to consider information from SEs in addition to genomic data; therefore, machine learning technology is expected to be introduced appropriately in terms of building a robust analysis platform with a high generalization performance. In this review, we explain the history and principles of SE, and the results of SE analysis using state-of-the-art machine learning and integrated analysis with other data are presented to provide a comprehensive understanding of the current status of SE analysis in the field of medical biology. Additionally, we compared the accuracy between existing machine learning methods on the benchmark dataset and attempted to explore the kind of data preprocessing and integration work needed to make the existing algorithms work on the benchmark dataset. Furthermore, we discuss the issues and future directions of current SE analysis.

show abstract

Residual Neural Network for Predicting Super-Enhancers on Genome Scale

Cited by 4 publications

References 28 publications

A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

Analysis of super-enhancer using machine learning and its application to medical biology

Contact Info

Product

Resources

About