List the 10% highest values of a distribution

Dear MatLab comunity,

I would like to obtain a list of values that are within 10% the highest value of the distribution that I have. The data is a list of aminoacids of a protein and their associated root mean square fluctuation.

My idea tough was to have not only the values of RMSF but also the residue associated with it in the list, so that I knew which value belongs to each residue.

So like:

[

Residue 234 3.3301

Residue 121 2.5150

...

et cetera

]

Here's the beginning of the script, I can get a list out of the 10% highest values but I need the association with the residue.

RMSF = load('rmsf.dat');

n = RMSF(:,1); %number of residue
x = RMSF(:,2); %RMSF for given residue

figure
hold on
title('RMSF UDP');
w1 = 0.5; 
bar(n,x,w1,'FaceColor','r')
xlabel('Residue');
ylabel('RMSF');
xlim([0 364])
ylim([0 5])

%%
Ms = sort(x_UDP,'descend');                          % Sort Descending
Result = Ms(1:ceil(length(Ms)*0.1))

Hope I have been somehow clear in stating the problem. Any suggestion about this would be highly appreciated!

All the best,

Alex

Answer 1

[Jan]

I guess that "x_UDP" is the same as "x".

RMSF = load('rmsf.dat');

n = RMSF(:,1); %number of residue
x = RMSF(:,2); %RMSF for given residue

[Ms, index] = sort(x, 'descend');
Result  = Ms(1:ceil(length(Ms)*0.1));
Result2 = n(index(1:ceil(length(Ms)*0.1)));

% Or easier:
RMSF_sorted = sortrows(RMSF, 2, 'descend');
Result      = RMSF_sorted(1:ceil(size(RMSF, 1) * 0.1), :)

[Alessandro Ruda]

Thank you very much for your reply! can ask you instead what would it be for the opposite case where I instead would like to list all the values within 10% of the lowest one?

In that case should I sort the list in the opposite order? Like:

RMSF_sorted = sortrows(RMSF, 2, 'ascend');
Result      = RMSF_sorted(1:ceil(size(RMSF, 1) * 0.1), :)

[Jan]

Yes.

Answer 2

[Veronica Taurino]

Change the last part:

[Ms, idx] = sort(x,'descend');                          % Sort Descending
Result = Ms(1:ceil(length(Ms)*0.1));
idx_result = idx(1:length(Result));