BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Short-Read DNA Sequence Alignment with Custom Designed FPGA-based 
 Hardware - Adam Hall\, University of Cambridge
DTSTART:20110304T160000Z
DTEND:20110304T170000Z
UID:TALK30099@talks.cam.ac.uk
CONTACT:Prof Simon Moore
DESCRIPTION:Short-read shotgun DNA sequencing is the determination of the 
 sequence\nof a large number of DNA reads which have been chemically cut ou
 t of the\ngenome sequence of a particular individual. Short-read DNA Seque
 ncing is\nthe underlying lab method which will be used for "personal genom
 ics."\nPersonal genomics is the tailoring of medical treatments for a\npar
 ticular individual. One area where personal genomics is likely to\nhave a 
 big impact is cancer treatment. It's likely to provide a more\naccurate an
 d automated method to map anti-cancer chemotherapy drugs to\nparticular ca
 ses of cancer than current methods. To tailor a treatment\nto a particular
  individual requires identifying the important\ndifferences between the st
 andard human genome sequence and the genome\nsequence of that particular i
 ndividual. Of the methods used for doing\nthis\, it appears that aligning 
 the short reads to the standard reference\ngenome is a necessary pre-proce
 ssing step.\n\nThe rate at which short read DNA sequence data is being pro
 duced doubles\nevery 5 months due to improvements in the hardware of the l
 aboratory\nequipment. As a result\, performing this alignment in a computa
 tionally\nefficient way is becoming increasingly important. We demonstrate
  how we\ncan exploit the ``embarrassingly parallel'' property of short rea
 d\nsequence alignment in custom-designed hardware in FPGA’s. Hardware is
 \nchosen\, a system is designed\, and this system is implemented. My\nFPGA
 -based hit finder was demonstrated to produce correct hit results.\nThe pe
 rformance of this single FPGA implementation was demonstrated to\nbe 71\,0
 00 seed hits found per hour on a human genome sized reference\nsequence. T
 he implementation was demonstrated to produce identical\nresults to the hi
 t finder stage of the widely-used software aligner MAQ.\nWe demonstrate th
 at the price/performance of this sliding-window FPGA\naligner (approximate
 ly 355 seeds/hr/$) compares favorably to the\nprice/performance of sliding
 -window software aligners (approximately\n67.5 seeds/hr/$ for MAQ). Howeve
 r\, software aligners which are based on\nthe superior Burrows-Wheeler ali
 gnment algorithm still have a\nsignificant price/performance advantage ove
 r the FPGA-based approach\n(approximately 7\,200 seeds/hr/$). We predict t
 hat as chips continue to\nincrease in size due to Moore’s Law and comput
 ation is performed in\nhigh-density cloud-computing datacenters the FPGA-b
 ased approach will\nbecome preferable to current software aligners.\n
LOCATION:SC04\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR