PSCOILS combines the sequence and the profile information using a linear
weighting scheme, namely
with
in the range of
the only variation with respect to the previous algorithm
is again the
equation (5 and 7). We then have
as before
![\begin{displaymath}
x_i=(\prod_{h=1}^W f(S,P,h,\lambda)^{e_h})^{1/N}
\end{displaymath}](img35.png) |
(8) |
and
![\begin{displaymath}
f(S,P,h,\lambda)=\lambda S^h(a_{i+h}) + (1-\lambda) <S^h,P_{i+h}>
\end{displaymath}](img36.png) |
(9) |
where the meaning of
and
are the same as in
5 and 7, respectively.
No attempt have been made to optimize
but in the current stage it has
been set to
. In this case the sequence and the profile are equally weighted.
Piero Fariselli
2008-04-06