27.12.2012 Views

l - People

l - People

l - People

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

!<br />

!<br />

!<br />

!<br />

!<br />

!<br />

!<br />

C = { StoneHinge,FO1,FO1M,FO,HingeSeq,TLSMD,NMA,NMB,NMC,NMD,1}<br />

x c (i) = output of predictor c for residue i.<br />

" c = weighting coefficient of predictor c, determined below.<br />

We used least squares fitting to find the " c‘s in Equation 5 corresponding to an optimal<br />

predictor. The procedure follows.<br />

Let<br />

y = a column vector, the components<br />

!<br />

!<br />

y(i) of which are the hinge annotations of the<br />

m residues in the HAG, in the format 1 = hinge, 0 = non-hinge. The index i counts over<br />

all residues in all proteins of the set in question, which in this work will be either the<br />

training, test, or complete HAG set. Order is unimportant as long as the i’s in<br />

the same order as the i’s in<br />

Let<br />

!<br />

x = an<br />

column of<br />

x, below.<br />

m " 9 matrix, the rows of which will be used to predict the ! rows of<br />

178<br />

!<br />

y are in<br />

y. Each<br />

x is a ! an m-component vector xc, such that c " C. Each component xc (i) of<br />

each such column vector is the output of the predictor c for residue i. Correspondingly,<br />

x(i) ! (without a subscript) is a row ! vector with ! 9 components corresponding ! to the output<br />

each of the 9 predictors emitted for residue i.<br />

Let " = a column vector, the components " c of which will give us the weight to be<br />

applied to the various predictors in order to make the composite HingeMaster predictor.<br />

Thus according to our definition of HingeMaster (Equation 5):<br />

!

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!