Biomolecular-Semiconductor Information Microsystems

我建议实现用于生物分子测量和分析的专用计算机芯片，特别是DNA测序。由于这些芯片将非常小，因为它们将消耗非常少的电力，它们将使DNA测量仪(“测序仪”)变得更便宜、更小、更便携。实际上，这项研究将有助于让更多的用户更容易获得和负担得起DNA测序，从而支持个性化医疗保健、生物分子环境传感等应用，甚至使用DNA本身的档案数据存储等下一代信息技术。 三种芯片被提出，都是为了一种测序形式，即单个DNA分子通过纳米级的孔(即所谓的纳米孔传感器)，这一行为会产生微小的DNA特征电流。对这一电流的分析揭示了DNA的组成，从而实现了测序。第一个定制芯片(模拟混合信号)负责放大和数字化从DNA传感纳米孔中出现的极弱的特征电流。第二个(数字)定制芯片处理这些数字化的测量信号，以计算被测DNA所组成的子分子(A、C、G、T)的序列，这一过程被称为“碱基连接”。第三个(数字)半定制芯片将来自第一个和第二个芯片的信号结合在一起，对组成所测量DNA的碱基序列做出更准确的预测，从而使结果对详细的临床分析有用。 这三个芯片中的每一个都旨在显著推进DNA信号处理的最先进水平。我对第一个芯片的研究重点是在不影响信号质量的情况下，将其速度提高至少10倍(这样它就可以处理更多的DNA测量)。我的研究还打算使第一个芯片可编程，并使其能够与纳米孔传感设备交互，以提高测序系统对制造和实验变化的适应性。我对第二个芯片的研究重点是将机器学习DNA序列分析推理算法集成到一个低于0.5W的DNA碱基加速器中。到目前为止，还没有这样的集成碱基计算器用于基于纳米孔的DNA测序仪。我对第三块芯片的研究集中在现场可编程门阵列(FPGA)硬件加速器上，该加速器与通用计算机配合使用，可以将DNA序列的校正速度提高100倍，同时所需的功耗大幅下降。到目前为止，还没有基于硬件的加速器用于基于纳米孔的测序校正的报道。

I propose the realization of specialized computer chips for biomolecule measurement and analysis, specifically DNA sequencing. Because these chips will be very small and because they will consume very little power, they shall allow DNA measurement machines ("sequencers") to become less costly and small and portable too. In effect, this research will help make DNA sequencing much more accessible and affordable to a broader set of users thus empowering applications such as personalized healthcare, biomolecular environmental sensing, and even next-generation information technologies such as archival data storage using DNA itself. Three chips are proposed, all for a form of sequencing whereby individual DNA molecules slip through a nano-scale aperture (a so-called “nanopore” sensor), an action that generates a minute signature-current of the DNA. An analysis of this current reveals the make-up of the DNA thus achieving sequencing. The first custom chip (analog mixed-signal) is responsible for amplifying and digitizing the extremely weak signature-currents emerging from the DNA-sensing nanopores. The second (digital) custom chip processes these digitized measurement signals to compute the sequence of sub-molecules (the "bases" A, C, G, T) that the measured DNA consists of, a process referred to as “basecalling”. The third (digital) semi-custom chip combines the signals from the first and second chip to make even more accurate predictions of the sequence of bases that make up the measured DNA, thus making the results useful for detailed clinical analysis. Each of these three chips aims to significantly advance the state-of-the-art in DNA signal processing. My research into the first chip is focused on extending its speed (so that it can handle more DNA measurements relative to competing approaches) by at least 10X without compromising signal quality. My research also intends to make the first chip programmable and will also enable it to interact with the nanopore sensing apparatus in order to improve the sequencing system's adaptability to manufacturing and experimental variations. My research into the second chip is focused on integrating machine-learning DNA sequence analysis inference algorithms into a sub-0.5 W DNA basecaller. To date, no such integrated basecaller exists for nanopore-based DNA sequencers. My research into the third chip is focused on a field-programmable-gate-array (FPGA) hardware accelerator to work in conjunction with a general-purpose computer to increase the rate at which DNA sequences can be corrected by 100X with a substantial drop in required power consumption. To date, no such hardware-based accelerator for nanopore-based sequencing correction has been reported.