| 1 |
# |
| 2 |
# This file contains a matrix used for weighing the training sequences |
| 3 |
# when given an input sequence. Let z = (da, dc, dg, dt) be the vector |
| 4 |
# containing the differences in the relative nucleotide frequencies of |
| 5 |
# two sequences, the input sequence and a training sequence. |
| 6 |
# Then the training sequence has weight proportional to |
| 7 |
# |
| 8 |
# exp ( - z M z^t) |
| 9 |
# |
| 10 |
# with M being the matrix specified below. |
| 11 |
# If M is nonsingular, then (apart from a two normalizing factors) M |
| 12 |
# is the inverse of the covariance matrix of a multinormal |
| 13 |
# distribution - the kernel for the estimation. |
| 14 |
|
| 15 |
|
| 16 |
# this matrix is gc-content only, i.e. |
| 17 |
# weight = 10 * exp (-200 * (dc + dg))^2) |
| 18 |
# in particular weight <= 10 |
| 19 |
0 0 0 0 |
| 20 |
0 200 0 0 |
| 21 |
0 0 200 0 |
| 22 |
0 0 0 0 |
| 23 |
|