Question

所有人！

我尝试在python中实现McCaskill算法（http://rna.informatik.uni-freiburg.de/Teaching/index.jsp?toolName=McCaskill)and在填充矩阵Q和Qbp时遇到了一些问题。我的代码是：

import sys
import numpy as np

def initialize_old(N, q = 0.):
    DP = np.empty((N,N))
    DP[:] = np.NAN
    for i in range(N):
        DP[i][i] = q
        if i != 0:
            DP[i][i-1] = q
    return DP

min_loop_length = 1 #Minimal loop length
RT = 1 #'Normalized' temperature 
Eij = -2 #Energy weight of base pair


Seq = 'GUUCC' #for instance consider such seq
L = len(Seq)
Q = initialize_old(L, 1.0)
Qbp = initialize_old(L)
p = initialize_old(L)
Q = np.c_[np.zeros(L),Q]
Q[0][0] = 1.0



Qbp = np.c_[np.zeros(L),Qbp]

def Eps_ij(i,j): 
    if (Seq[i].upper(),Seq[j].upper()) in (('A','U'),('U','A'),('C','G'),('G','C'),('G','U')):
        Qbp[i][j] = Q[i+1][j-1] * float(np.exp(-Eij / RT))
        return float(Qbp[i][j])
    return 0.0

def McCaskill():
    for k in range(1, L+1):
        for i in range(L-k+1):
            j = i + k -1

            sum = 0.0
            for h in range(i,j-1):
                sum += Q[i][h-1] * Eps_ij(h,j)


            Q[i][j] = Q[i][j-1] + sum

if __name__ == '__main__':

    McCaskill()

    print(Q)
    print(Qbp)

输出为：

Q = [[  1.   1.  nan  nan  nan  nan]
 [  0.   1.   1.   1.   1.  nan]
 [  0.  nan   1.   1.  nan  nan]
 [  0.  nan  nan   1.   1.  nan]
 [  0.  nan  nan  nan   1.   1.]]

Qbp = [[ 0.         0.         7.3890561  7.3890561  7.3890561        nan]
 [ 0.         0.         0.               nan        nan        nan]
 [ 0.               nan  0.         0.               nan        nan]
 [ 0.               nan        nan  0.         0.               nan]
 [ 0.               nan        nan        nan  0.         0.       ]]

由于缺少初始条件和递归范围的详细信息，我无法理解如何逐步填充两个数组。

非常感谢您提供任何建议和想法！

谢谢！

最诚挚的问候！

用于RNA二级结构预测的McCaskill算法的Python实现。

0 个答案: