how to get rid of the too many if statements..

Question

Jwana 2012-11-11

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/53434-how-to-get-rid-of-the-too-many-if-statements

I have 2 cell matrix (coming from a certain code):

   m1 = 
      'GO:0008150'    'GO:0016740'
      'GO:0016740'    'GO:0016787'
      'GO:0016787'    'GO:0006810'
      'GO:0008150'    'GO:0006412'
      'GO:0016740'    'GO:0004672'
      'GO:0016740'    'GO:0016779'
      'GO:0016787'    'GO:0004386'
      'GO:0016787'    'GO:0003774'
      'GO:0016787'    'GO:0016298'
      'GO:0006810'    'GO:0016192'
m2 = 
    'GO:0008150'    'GO:0016787'
    'GO:0008150'    'GO:0006810'
    'GO:0006810'    'GO:0006412'
    'GO:0016192'    'GO:0003774'
    'GO:0006810'    'GO:0005215'
    'GO:0005215'    'GO:0030533'

in each matrix, the first column is the parent of the second one... I need from the code to find the part of graph which can determine all relations between these cells... I have to find cell from column 1 that occurance in column 2, then take the cell from column 1 but from the same row of column 2 and find the similarity between it and onother cell in column 2 .. and so.. it is something like depth first search in graph theory

In general, How can I make a dynamic code (recursive) that can replace the following if statements:

for k=1:length(m1)
for ii=1:length(m1)         
for j=1:length(m1)
for i=1:length(m2)
 for e=1:length(m1)
   if isequal(m2{i,1},m1{j,2})
      x1=[m1(j,1) m2(i,2)];
      x11=[x1;x11];
       if isequal(m1{j,1},m1{k,2})
          x2=[m1(k,1),m2(i,2)];
          x22=[x2;x22];
             if isequal(m1{k,1},m1{ii,2})
              x3=[m1(ii,1),m2(i,2)];
              x33=[x3;x33];
                 if isequal(m1{ii,1},m1{e,2})
                 x4=[m1(e,1),m2(i,2)];
                 x44=[x4;x44];
                      .
                      .
            and so..
x_total=[x11;x22;x33;x44...]
end
end
end
end
end
end
end

Note that the if statemtnes are not determined( it depends on m1 & m2 which are also not always constants) I think I have to use recursion technique..

5 个评论
显示 3更早的评论隐藏 3更早的评论

Arthur 2012-11-11

在 MATLAB Online 中打开

Why do you loop 4 times over the same array? Every if statement does exactly the same thing. What do you want to achieve with this script???

Note not that you can also use strcmp to find matching strings in a cell array. For instance:

idx = strcmp(m1{1},m1);

will find you all values matching m1{1}.

Jwana 2012-11-12

The first column is the parent of the second one... I need from these if statements to find the part of graph which can determine all relations between these cells... it is not the same value every time you search in m1 cells array.. you have to find cell from column 1 that occurance in column 2, then the take the cell from column 1 but from the same row of column 2 and find the similarity between it and ontother cell in column 2 .. and so..

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2012-11-11

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/53434-how-to-get-rid-of-the-too-many-if-statements#answer_65040

编辑：Star Strider 2012-11-11

在 MATLAB Online 中打开

I am not certain what you want to do, so I am guessing here.

Would something like this do what you want:

[C ia ic] = intersect( m1(:,1), m1(:,2) );

The intersect function looks for occurrences common to both columns in your m1 array. you then only need to look for occurrences of the same index in ia and ic to find values that are common to both columns in the same rows.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Jwana 2012-11-13

Thank you for your response... The first column is the parent of the second one... I need from these if statements to find the part of graph which can determine all relations between these cells... it is not the same value every time you search in m1 cells array.. you have to find cell from column 1 that occurance in column 2, then the take the cell from column 1 but from the same row of column 2 and find the similarity between it and ontother cell in column 2 .. and so..

请先登录，再进行评论。