Thursday, January 16, 2014, 11:00am - 12:00pm
Associate Professor of ECE at Iowa State University
"Maximum Likelihood Sequence Detection: From DNA Sequencing to Nano-Imaging"
Abstract: The basic problem of inferring useful information from a sequence of noisy data appears in several domains such as communications, speech processing and radar signal processing. Such problems are often solved via system modeling and optimal sequence detection techniques and have revolutionized technologies such as hard drives, cellphones and embedded speech recognition engines. In this talk I will argue that a systems viewpoint allows us to leverage the power of optimal sequence detection techniques in the seemingly disparate fields of DNA sequencing and atomic force microscope (AFM) based nano-imaging. Shotgun DNA sequencing operates by randomly fragmenting a genome into reads that are a few hundred of base pairs long. Current sequencers produce reads at a high throughput; however, they are often corrupted by errors which pose a serious problem for downstream genome assembly. Correcting errors in reads is thus an important problem. The problem differs significantly from classical error correction since there is a lack of alignment information between different reads. I will present our recent algorithm for the error correction of high throughput DNA sequencing data that relies on maximum likelihood sequence detection. Our test results on publicly available C. elegans and E. coli datasets show that our algorithm achieves significantly better performance than other competing methods in the bioinformatics domain. A completely different application where sequence detection gives huge gains is that of dynamic mode AFM imaging. An AFM is a device where a sharp cantilever probe (few nanometers radius) is used to sense atomic level images. In this domain the key challenge is to appropriately infer the image from minute changes in the cantilever deflection signal. The dynamic mode is the preferred mode of imaging for soft samples such as biological cells. However, using existing techniques it is quite slow. I will present our work in this area that operates by viewing the system as a communication channel and reformulating topographic feature detection as an optimal sequence detection problem. Experimental results demonstrate that our detector is much faster than conventional techniques. This talk will be self-contained. No background in DNA sequencing or AFM will be assumed.
Bio: Aditya Ramamoorthy is an Associate Professor of Electrical and Computer Engineering at Iowa State University. He received his B. Tech. degree in Electrical Engineering from the Indian Institute of Technology, Delhi in 1999 and the M.S. and Ph.D. degrees from the University of California, Los Angeles (UCLA) in 2002 and 2005 respectively. He was a systems engineer at Biomorphic VLSI Inc. till 2001. From 2005 to 2006 he was with the data storage signal processing group at Marvell Semiconductor Inc. His research interests are in the areas of network information theory, channel coding and signal processing for nanotechnology and bioinformatics. Dr. Ramamoorthy has been serving as an associate editor for the IEEE Transactions on Communications since Nov. 2011. He is the recipient of the 2012 Iowa State University's Early Career Engineering Faculty Research Award, the 2012 NSF CAREER award, and the Harpole-Pentair professorship in 2009 and 2010.
Hosted by: Prof. Dongning Guo and EECS S&S Division