BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Smoothed absolute loadings principal components analysis - Bernie 
 Silverman (Oxford)
DTSTART:20100305T153000Z
DTEND:20100305T163000Z
UID:TALK22582@talks.cam.ac.uk
CONTACT:8047
DESCRIPTION:A crucial part of genome-wide association studies is the\niden
 tification of modes of variability in genome data which do not\ndepend on 
 small parts of the genome.  The natural statistical\nstarting-point is pri
 ncipal components analysis\, but in practice raw\nprincipal components pro
 duce loadings concentrated on a small number\nof SNPs. Therefore some sort
  of regularization is required.\n\nStandard Functional Data Analysis appro
 aches control the amount of\nlocal variability in the loadings vector\, bu
 t this is not appropriate\nin the current case\, because of the arbitrary 
 coding of the individual\nSNPs. Therefore a regularization method for the 
 absolute values of the\nloadings is developed and discussed.  Interestingl
 y\, a promising\ncomputational approach within the method is Lamarckian ge
 netic\nalgorithms\, thus illustrating the remark in the literature that\n"
 Lamarckism has been universally rejected as a viable theory of\ngenetic ev
 olution in nature but Lamarckian evolution has proven\neffective within co
 mputer applications"!\n\nhttp://www.stats.ox.ac.uk/~silverma/
LOCATION:MR12\, CMS\, Wilberforce Road\, Cambridge\, CB3 0WB
END:VEVENT
END:VCALENDAR