A Framework for Designing Efficient Deep Learning-Based Genomic Basecallers

TL;DR: Zhang et al. as mentioned in this paper proposed a quantization-aware base calling neural architecture search (QABAS) framework to find the best bit-width precision for each neural network layer.

Abstract: Nanopore sequencing is a widely-used high-throughput genome sequencing technology that can se-quence long fragments of a genome. Nanopore sequencing generates noisy electrical signals that need to be converted into a standard string of DNA nucleotide bases using a computational step called basecalling. The performance of basecalling has critical implications for all later steps in genome analysis. Many researchers adopt complex deep learning-based models from the speech recognition domain to perform basecalling without considering the compute demands of such models, which leads to slow, inefficient, and memory-hungry basecallers. Therefore, there is a need to reduce the computation and memory cost of basecalling while maintaining accuracy. However, developing a very fast basecaller that can provide high accuracy requires a deep understanding of genome sequencing, machine learning, and hardware design. Our goal is to develop a comprehensive framework for creating deep learning-based basecallers that provide high efficiency and performance. 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 first 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 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 RUBICON 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, memory-efficient, and hardware-friendly basecaller. Compared to the fastest state-of-the-art basecaller, RUBICALL provides a 3.96× speedup with 2.97% higher accuracy. Compared to an expert-designed basecaller, RUBICALL provides a 141.15× 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. We show that RUBICON helps researchers develop hardware-optimized basecallers that are superior to expert-designed models and can inspire independent future ideas.