max/ourMELONS
1
0
Fork 0
Code Issues 1 Pull requests Projects Releases Packages Wiki Activity Actions

Added source Matlab code for reference

Browse source
This commit is contained in:
Waldir Leoncio 2019-12-16 16:47:21 +01:00
parent b8af977117
commit b5d99903d2
186 changed files with 61405 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

776
matlab/parallel/admix_parallel.m Normal file
View file

@ -0,0 +1,776 @@
function admix_parallel(options)
% ADMIX_PARALLEL is the command line version of the baps partition with
% admixture models.
% Input: options is a struct generated by parallel.m
%--------------------------------------------------------------------------
%- Syntax check out
%--------------------------------------------------------------------------
rand('state',0); % used for debugging
outp = [options.outputMat '.txt'];
inp = options.dataFile;
clusters = options.clusters;
fprintf(1,'Parallel computing...\n');
fprintf(1,'Admixture analysis for cluster(s): %s.\n',num2str(clusters));
global COUNTS; global PARTITION; global SUMCOUNTS;
clearGlobalVars;
struct_array = load(options.dataFile);
if isfield(struct_array,'c') %Matlab versio
c = struct_array.c;
if ~isfield(c,'PARTITION') | ~isfield(c,'rowsFromInd')
disp('*** ERROR: Incorrect data format');
return
end
elseif isfield(struct_array,'PARTITION') %Mideva versio
c = struct_array;
if ~isfield(c,'rowsFromInd')
disp('*** ERROR: Incorrect data format');
return
end
else
disp('*** ERROR: Incorrect data format');
return;
end
if isfield(c, 'gene_lengths') && ...
(strcmp(c.mixtureType,'linear_mix') | ...
strcmp(c.mixtureType,'codon_mix')) % if the mixture is from a linkage model
% Redirect the call to the linkage admixture function.
fprintf(1,'Redirecting to Linkage Model Admixture\n');
c.data = noIndex(c.data,c.noalle); % call function noindex to remove the index column
linkage_admix_parallel(c,options);
return
end
% This section is disabled, -Jing 27.10.2009
% if isfield(c, 'gene_lengths') && ...
% (strcmp(c.mixtureType,'linkage_mix') | ...
% strcmp(c.mixtureType,'codon_mix')) % if the mixture is from a linkage model
% % Redirect the call to the linkage admixture function.
% c.data = noIndex(c.data,c.noalle); % call function noindex to remove the index column
% linkage_admix(c);
% return
% end
PARTITION = c.PARTITION; COUNTS = c.COUNTS; SUMCOUNTS = c.SUMCOUNTS;
alleleCodes = c.alleleCodes; adjprior = c.adjprior; popnames = c.popnames;
rowsFromInd = c.rowsFromInd; data = c.data; npops = c.npops; noalle = c.noalle;
% answers = inputdlg({['Input the minimum size of a population that will'...
% ' be taken into account when admixture is estimated.']},...
% 'Input minimum population size',[1],...
% {'5'});
% if isempty(answers) return; end
% -------------------------------------------
% NEW: for parallel computing
% -------------------------------------------
alaRaja = options.minSize;
[npops, clusterIndex] = poistaLiianPienet(npops, rowsFromInd, alaRaja);
if length(clusterIndex)<length(clusters)
disp('*** ERROR: error in cluster labels.');
return
end
clusters = clusterIndex(clusters); % after removing outlier clusters
nloci = size(COUNTS,2);
ninds = size(data,1)/rowsFromInd;
% answers = inputdlg({['Input number of iterations']},'Input',[1],{'50'});
% if isempty(answers) return; end
iterationCount = options.iters;
% answers = inputdlg({['Input number of reference individuals from each population']},'Input',[1],{'50'});
% if isempty(answers) nrefIndsInPop = 50;
% else nrefIndsInPop = str2num(answers{1,1});
% end
nrefIndsInPop = options.refInds;
% answers = inputdlg({['Input number of iterations for reference individuals']},'Input',[1],{'10'});
% if isempty(answers) return; end
% iterationCountRef = str2num(answers{1,1});
iterationCountRef = options.refIters;
% First calculate log-likelihood ratio for all individuals:
likelihood = zeros(ninds,1);
allfreqs = computeAllFreqs2(noalle);
for ind = 1:ninds
omaFreqs = computePersonalAllFreqs(ind, data, allfreqs, rowsFromInd);
osuusTaulu = zeros(1,npops);
if PARTITION(ind)==0
% Yksil?on outlier
elseif PARTITION(ind)~=0
if PARTITION(ind)>0
osuusTaulu(PARTITION(ind)) = 1;
else
% Yksilöt, joita ei ole sijoitettu mihinkään koriin.
arvot = zeros(1,npops);
for q=1:npops
osuusTaulu = zeros(1,npops);
osuusTaulu(q) = 1;
arvot(q) = computeIndLogml(omaFreqs, osuusTaulu);
end
[iso_arvo, isoimman_indeksi] = max(arvot);
osuusTaulu = zeros(1,npops);
osuusTaulu(isoimman_indeksi) = 1;
PARTITION(ind)=isoimman_indeksi;
end
logml = computeIndLogml(omaFreqs, osuusTaulu);
logmlAlku = logml;
for osuus = [0.5 0.25 0.05 0.01]
[osuusTaulu, logml] = etsiParas(osuus, osuusTaulu, omaFreqs, logml);
end
logmlLoppu = logml;
likelihood(ind) = logmlLoppu-logmlAlku;
end
end
% Analyze further only individuals who have log-likelihood ratio larger than 3:
% ---------------------------------------
% PARALLEL COMPUTING IMPLEMENTED HERE
% ---------------------------------------
to_investigate = (find(likelihood>3))';
admix_populaatiot = unique(PARTITION(to_investigate));
validCluster = intersect(clusters, admix_populaatiot); % for the chosen clusters
ix = zeros(length(to_investigate),1);
for i = 1:length(validCluster)
ix = ix | (PARTITION(to_investigate) == validCluster(i));
end
if ~any(ix) to_investigate = [];
else to_investigate = to_investigate(ix);
end
admix_populaatiot = unique(PARTITION(to_investigate));
disp('Possibly admixed individuals in the chosen clusters: ');
if isempty(to_investigate)
disp('none');
disp('Admixture analysis terminated.');
return
else
for i = 1:length(to_investigate)
disp(num2str(to_investigate(i)));
end
end
disp(' ');
disp('clusters for possibly admixed individuals: ');
for i = 1:length(admix_populaatiot)
disp(num2str(admix_populaatiot(i)));
end
% THUS, there are two types of individuals, who will not be analyzed with
% simulated allele frequencies: those who belonged to a mini-population
% which was removed, and those who have log-likelihood ratio less than 3.
% The value in the PARTITION for the first kind of individuals is 0. The
% second kind of individuals can be identified, because they do not
% belong to "to_investigate" array. When the results are presented, the
% first kind of individuals are omitted completely, while the second kind
% of individuals are completely put to the population, where they ended up
% in the mixture analysis. These second type of individuals will have a
% unit p-value.
% Simulate allele frequencies a given number of times and save the average
% result to "proportionsIt" array.
proportionsIt = zeros(ninds,npops);
for iterationNum = 1:iterationCount
disp(['Iter: ' num2str(iterationNum)]);
allfreqs = simulateAllFreqs(noalle); % Allele frequencies on this iteration.
for ind=to_investigate
%disp(num2str(ind));
omaFreqs = computePersonalAllFreqs(ind, data, allfreqs, rowsFromInd);
osuusTaulu = zeros(1,npops);
if PARTITION(ind)==0
% Yksil?on outlier
elseif PARTITION(ind)~=0
if PARTITION(ind)>0
osuusTaulu(PARTITION(ind)) = 1;
else
% Yksilöt, joita ei ole sijoitettu mihinkään koriin.
arvot = zeros(1,npops);
for q=1:npops
osuusTaulu = zeros(1,npops);
osuusTaulu(q) = 1;
arvot(q) = computeIndLogml(omaFreqs, osuusTaulu);
end
[iso_arvo, isoimman_indeksi] = max(arvot);
osuusTaulu = zeros(1,npops);
osuusTaulu(isoimman_indeksi) = 1;
PARTITION(ind)=isoimman_indeksi;
end
logml = computeIndLogml(omaFreqs, osuusTaulu);
for osuus = [0.5 0.25 0.05 0.01]
[osuusTaulu, logml] = etsiParas(osuus, osuusTaulu, omaFreqs, logml);
end
end
proportionsIt(ind,:) = proportionsIt(ind,:).*(iterationNum-1) + osuusTaulu;
proportionsIt(ind,:) = proportionsIt(ind,:)./iterationNum;
end
end
%disp(['Creating ' num2str(nrefIndsInPop) ' reference individuals from ']);
%disp('each population.');
%allfreqs = simulateAllFreqs(noalle); % Simuloidaan alleelifrekvenssisetti
allfreqs = computeAllFreqs2(noalle); % Koitetaan tällaista.
% Initialize the data structures, which are required in taking the missing
% data into account:
n_missing_levels = zeros(npops,1); % number of different levels of "missingness" in each pop (max 3).
missing_levels = zeros(npops,3); % the mean values for different levels.
missing_level_partition = zeros(ninds,1); % level of each individual (one of the levels of its population).
for i=1:npops
inds = find(PARTITION==i);
% Proportions of non-missing data for the individuals:
non_missing_data = zeros(length(inds),1);
for j = 1:length(inds)
ind = inds(j);
non_missing_data(j) = length(find(data((ind-1)*rowsFromInd+1:ind*rowsFromInd,:)>0)) ./ (rowsFromInd*nloci);
end
if all(non_missing_data>0.9)
n_missing_levels(i) = 1;
missing_levels(i,1) = mean(non_missing_data);
missing_level_partition(inds) = 1;
else
[ordered, ordering] = sort(non_missing_data);
%part = learn_simple_partition(ordered, 0.05);
part = learn_partition_modified(ordered);
aux = sortrows([part ordering],2);
part = aux(:,1);
missing_level_partition(inds) = part;
n_levels = length(unique(part));
n_missing_levels(i) = n_levels;
for j=1:n_levels
missing_levels(i,j) = mean(non_missing_data(find(part==j)));
end
end
end
% Create and analyse reference individuals for populations
% with potentially admixed individuals:
refTaulu = zeros(npops,100,3);
for pop = admix_populaatiot'
for level = 1:n_missing_levels(pop)
potential_inds_in_this_pop_and_level = ...
find(PARTITION==pop & missing_level_partition==level &...
likelihood>3); % Potential admix individuals here.
if ~isempty(potential_inds_in_this_pop_and_level)
%refData = simulateIndividuals(nrefIndsInPop,rowsFromInd,allfreqs);
refData = simulateIndividuals(nrefIndsInPop, rowsFromInd, allfreqs, ...
pop, missing_levels(pop,level));
disp(['Analysing the reference individuals from pop ' num2str(pop) ' (level ' num2str(level) ').']);
refProportions = zeros(nrefIndsInPop,npops);
for iter = 1:iterationCountRef
%disp(['Iter: ' num2str(iter)]);
allfreqs = simulateAllFreqs(noalle);
for ind = 1:nrefIndsInPop
omaFreqs = computePersonalAllFreqs(ind, refData, allfreqs, rowsFromInd);
osuusTaulu = zeros(1,npops);
osuusTaulu(pop)=1;
logml = computeIndLogml(omaFreqs, osuusTaulu);
for osuus = [0.5 0.25 0.05 0.01]
[osuusTaulu, logml] = etsiParas(osuus, osuusTaulu, omaFreqs, logml);
end
refProportions(ind,:) = refProportions(ind,:).*(iter-1) + osuusTaulu;
refProportions(ind,:) = refProportions(ind,:)./iter;
end
end
for ind = 1:nrefIndsInPop
omanOsuus = refProportions(ind,pop);
if round(omanOsuus*100)==0
omanOsuus = 0.01;
end
if abs(omanOsuus)<1e-5
omanOsuus = 0.01;
end
refTaulu(pop, round(omanOsuus*100),level) = refTaulu(pop, round(omanOsuus*100),level)+1;
end
end
end
end
% Rounding of the results:
proportionsIt = proportionsIt.*100; proportionsIt = round(proportionsIt);
proportionsIt = proportionsIt./100;
for ind = 1:ninds
if ~any(to_investigate==ind)
if PARTITION(ind)>0
proportionsIt(ind,PARTITION(ind))=1;
end
else
% In case of a rounding error, the sum is made equal to unity by
% fixing the largest value.
if (PARTITION(ind)>0) & (sum(proportionsIt(ind,:)) ~= 1)
[isoin,indeksi] = max(proportionsIt(ind,:));
erotus = sum(proportionsIt(ind,:))-1;
proportionsIt(ind,indeksi) = isoin-erotus;
end
end
end
% Calculate p-value for each individual:
uskottavuus = zeros(ninds,1);
for ind = 1:ninds
pop = PARTITION(ind);
if pop==0 % Individual is outlier
uskottavuus(ind)=1;
elseif isempty(find(to_investigate==ind))
% Individual had log-likelihood ratio<3
uskottavuus(ind)=1;
else
omanOsuus = proportionsIt(ind,pop);
if abs(omanOsuus)<1e-5
omanOsuus = 0.01;
end
if round(omanOsuus*100)==0
omanOsuus = 0.01;
end
level = missing_level_partition(ind);
refPienempia = sum(refTaulu(pop, 1:round(100*omanOsuus), level));
uskottavuus(ind) = refPienempia / nrefIndsInPop;
end
end
% tulostaAdmixtureTiedot(proportionsIt, uskottavuus, alaRaja, iterationCount);
% viewPartition(proportionsIt, popnames);
c.proportionsIt = proportionsIt;
c.pvalue = uskottavuus; % Added by Jing
c.mixtureType = 'admix'; % Jing
c.admixnpops = npops;
c.clusters = clusters; % added for parallel computing
c.minsize = alaRaja;
c.iters = iterationCount;
c.refInds = nrefIndsInPop;
c.refIters = iterationCountRef;
fprintf(1,'Saving the result...')
try
% save(options.outputMat, 'c');
save(options.outputMat, 'c','-v7.3'); % added by Lu Cheng, 08.06.2012
fprintf(1,'Finished.\n');
catch
display('*** ERROR in saving the result.');
end
%----------------------------------------------------------------------------
function [npops, clusterIndex] = poistaLiianPienet(npops, rowsFromInd, alaraja)
% Muokkaa tulokset muotoon, jossa outlier yksilöt on
% poistettu. Tarkalleen ottaen poistaa ne populaatiot,
% joissa on vähemmän kuin 'alaraja':n verran yksilöit?
% NEW: clusterIndex is output for parallel computing. - Jing
global PARTITION;
global COUNTS;
global SUMCOUNTS;
clusterIndex = [1:npops]';
popSize=zeros(1,npops);
for i=1:npops
popSize(i)=length(find(PARTITION==i));
end
miniPops = find(popSize<alaraja);
if length(miniPops)==0
return;
end
outliers = [];
for pop = miniPops
inds = find(PARTITION==pop);
disp('Removed individuals: ');
disp(num2str(inds));
outliers = [outliers; inds];
clusterIndex(pop) = 0;
end
ninds = length(PARTITION);
PARTITION(outliers) = 0;
korit = unique(PARTITION(find(PARTITION>0)));
for n=1:length(korit)
kori = korit(n);
yksilot = find(PARTITION==kori);
PARTITION(yksilot) = n;
clusterIndex(kori) = n;
end
COUNTS(:,:,miniPops) = [];
SUMCOUNTS(miniPops,:) = [];
npops = npops-length(miniPops);
%------------------------------------------------------------------------
function clearGlobalVars
global COUNTS; COUNTS = [];
global SUMCOUNTS; SUMCOUNTS = [];
global PARTITION; PARTITION = [];
global POP_LOGML; POP_LOGML = [];
%--------------------------------------------------------
function allFreqs = computeAllFreqs2(noalle)
% Lisää a priori jokaista alleelia
% joka populaation joka lokukseen j 1/noalle(j) verran.
global COUNTS;
global SUMCOUNTS;
max_noalle = size(COUNTS,1);
nloci = size(COUNTS,2);
npops = size(COUNTS,3);
sumCounts = SUMCOUNTS+ones(size(SUMCOUNTS));
sumCounts = reshape(sumCounts', [1, nloci, npops]);
sumCounts = repmat(sumCounts, [max_noalle, 1 1]);
prioriAlleelit = zeros(max_noalle,nloci);
for j=1:nloci
prioriAlleelit(1:noalle(j),j) = 1/noalle(j);
end
prioriAlleelit = repmat(prioriAlleelit, [1,1,npops]);
counts = COUNTS + prioriAlleelit;
allFreqs = counts./sumCounts;
function allfreqs = simulateAllFreqs(noalle)
% Lisää jokaista alleelia joka populaation joka lokukseen j 1/noalle(j)
% verran. Näin saatuja counts:eja vastaavista Dirichlet-jakaumista
% simuloidaan arvot populaatioiden alleelifrekvensseille.
global COUNTS;
max_noalle = size(COUNTS,1);
nloci = size(COUNTS,2);
npops = size(COUNTS,3);
prioriAlleelit = zeros(max_noalle,nloci);
for j=1:nloci
prioriAlleelit(1:noalle(j),j) = 1/noalle(j);
end
prioriAlleelit = repmat(prioriAlleelit, [1,1,npops]);
counts = COUNTS + prioriAlleelit;
allfreqs = zeros(size(counts));
for i=1:npops
for j=1:nloci
simuloidut = randdir(counts(1:noalle(j),j,i) , noalle(j));
allfreqs(1:noalle(j),j,i) = simuloidut;
end
end
%--------------------------------------------------------------------------
function refData = simulateIndividuals(n,rowsFromInd,allfreqs,pop, missing_level)
% simulate n individuals from population pop, such that approximately
% proportion "missing_level" of the alleles are present.
nloci = size(allfreqs,2);
refData = zeros(n*rowsFromInd,nloci);
counter = 1; % which row will be generated next.
for ind = 1:n
for loc = 1:nloci
for k=0:rowsFromInd-1
if rand<missing_level
refData(counter+k,loc) = simuloiAlleeli(allfreqs,pop,loc);
else
refData(counter+k,loc) = -999;
end
end
end
counter = counter+rowsFromInd;
end
function all = simuloiAlleeli(allfreqs,pop,loc)
% Simuloi populaation pop lokukseen loc alleelin.
freqs = allfreqs(:,loc,pop);
cumsumma = cumsum(freqs);
arvo = rand;
isommat = find(cumsumma>arvo);
all = min(isommat);
%--------------------------------------------------------------------------
function omaFreqs = computePersonalAllFreqs(ind, data, allFreqs, rowsFromInd)
% Laskee npops*(rowsFromInd*nloci) taulukon, jonka kutakin saraketta
% vastaa yksilön ind alleeli. Eri rivit ovat alleelin alkuperäfrekvenssit
% eri populaatioissa. Jos yksilölt?puuttuu jokin alleeli, niin vastaavaan
% kohtaa tulee sarake ykkösi?
global COUNTS;
nloci = size(COUNTS,2);
npops = size(COUNTS,3);
rows = data(computeRows(rowsFromInd, ind, 1),:);
omaFreqs = zeros(npops, (rowsFromInd*nloci));
pointer = 1;
for loc=1:size(rows,2)
for all=1:size(rows,1)
if rows(all,loc)>=0
try,
omaFreqs(:,pointer) = ...
reshape(allFreqs(rows(all,loc),loc,:), [npops,1]);
catch
a=0;
end
else
omaFreqs(:,pointer) = ones(npops,1);
end
pointer = pointer+1;
end
end
%---------------------------------------------------------------------------
function loggis = computeIndLogml(omaFreqs, osuusTaulu)
% Palauttaa yksilön logml:n, kun oletetaan yksilön alkuperät
% määritellyiksi kuten osuusTaulu:ssa.
apu = repmat(osuusTaulu', [1 size(omaFreqs,2)]);
apu = apu .* omaFreqs;
apu = sum(apu);
apu = log(apu);
loggis = sum(apu);
%--------------------------------------------------------------------------
function osuusTaulu = suoritaMuutos(osuusTaulu, osuus, indeksi)
% Päivittää osuusTaulun muutoksen jälkeen.
global COUNTS;
npops = size(COUNTS,3);
i1 = rem(indeksi,npops);
if i1==0, i1 = npops; end;
i2 = ceil(indeksi / npops);
osuusTaulu(i1) = osuusTaulu(i1)-osuus;
osuusTaulu(i2) = osuusTaulu(i2)+osuus;
%-------------------------------------------------------------------------
function [osuusTaulu, logml] = etsiParas(osuus, osuusTaulu, omaFreqs, logml)
ready = 0;
while ready ~= 1
muutokset = laskeMuutokset4(osuus, osuusTaulu, omaFreqs, logml);
[maxMuutos, indeksi] = max(muutokset(1:end));
if maxMuutos>0
osuusTaulu = suoritaMuutos(osuusTaulu, osuus, indeksi);
logml = logml + maxMuutos;
else
ready = 1;
end
end
%---------------------------------------------------------------------------
function muutokset = laskeMuutokset4(osuus, osuusTaulu, omaFreqs, logml)
% Palauttaa npops*npops taulun, jonka alkio (i,j) kertoo, mik?on
% muutos logml:ss? mikäli populaatiosta i siirretään osuuden verran
% todennäköisyysmassaa populaatioon j. Mikäli populaatiossa i ei ole
% mitään siirrettävää, on vastaavassa kohdassa rivi nollia.
global COUNTS;
npops = size(COUNTS,3);
notEmpty = find(osuusTaulu>0.005);
muutokset = zeros(npops);
empties = ~notEmpty;
for i1=notEmpty
osuusTaulu(i1) = osuusTaulu(i1)-osuus;
for i2 = [1:i1-1 i1+1:npops]
osuusTaulu(i2) = osuusTaulu(i2)+osuus;
loggis = computeIndLogml(omaFreqs, osuusTaulu);
muutokset(i1,i2) = loggis-logml;
osuusTaulu(i2) = osuusTaulu(i2)-osuus;
end
osuusTaulu(i1) = osuusTaulu(i1)+osuus;
end
function g=randga(a,b)
flag = 0;
if a>1
c1 = a-1; c2 = (a-(1/(6*a)))/c1; c3 = 2/c1; c4 = c3+2; c5 = 1/sqrt(a);
U1=-1;
while flag == 0,
if a<=2.5,
U1=rand;U2=rand;
else
while ~(U1>0 & U1<1),
U1=rand;U2=rand;
U1 = U2 + c5*(1-1.86*U1);
end %while
end %if
W = c2*U2/U1;
if c3*U1+W+(1/W)<=c4,
flag = 1;
g = c1*W/b;
elseif c3*log(U1)-log(W)+W<1,
flag = 1;
g = c1*W/b;
else
U1=-1;
end %if
end %while flag
elseif a==1
g=sum(-(1/b)*log(rand(a,1)));
else
while flag == 0,
U = rand(2,1);
if U(1)>exp(1)/(a+exp(1)),
g = -log(((a+exp(1))*(1-U(1)))/(a*exp(1)));
if U(2)<=g^(a-1),
flag = 1;
end %if
else
g = ((a+exp(1))*U(1)/((exp(1))^(1/a)));
if U(2)<=exp(-g),
flag = 1;
end %if
end %if
end %while flag
g=g/b;
end %if;
%-------------------------------------------------
function svar=randdir(counts,nc)
% Käyttöesim randdir([10;30;60],3)
svar=zeros(nc,1);
for i=1:nc
svar(i,1)=randga(counts(i,1),1);
end
svar=svar/sum(svar);
%-------------------------------------------------------------------------------------
function rows = computeRows(rowsFromInd, inds, ninds)
% Individuals inds have been given. The function returns a vector,
% containing the indices of the rows, which contain data from the
% individuals.
rows = inds(:, ones(1,rowsFromInd));
rows = rows*rowsFromInd;
miinus = repmat(rowsFromInd-1 : -1 : 0, [ninds 1]);
rows = rows - miinus;
rows = reshape(rows', [1,rowsFromInd*ninds]);
%--------------------------------------------------------------------------
%-----
function str = ownNum2Str(number)
absolute = abs(number);
if absolute < 1000
str = num2str(number);
elseif absolute < 10000000
first_three = rem(number,1000);
next_four = (number - first_three) /1000;
first_three = abs(first_three);
if first_three<10
first_three = ['00' num2str(first_three)];
elseif first_three<100
first_three = ['0' num2str(first_three)];
else
first_three = num2str(first_three);
end;
str = [num2str(next_four) first_three];
elseif absolute < 100000000
first_four = rem(number,10000);
next_four = (number - first_four) /10000;
first_four = abs(first_four);
if first_four<10
first_four = ['000' num2str(first_four)];
elseif first_four<100
first_four = ['00' num2str(first_four)];
elseif first_four<1000
first_four = ['0' num2str(first_four)];
else
first_four = num2str(first_four);
end;
str = [num2str(next_four) first_four];
else
str = num2str(number);
end;
%------------------------------------------------
function part = learn_partition_modified(ordered)
% This function is called only if some individual has less than 90 per cent
% non-missing data. The function uses fuzzy clustering for the "non-missingness"
% values, finding maximum three clusters. If two of the found clusters are such
% that all the values are >0.9, then those two are further combined.
part = learn_simple_partition(ordered,0.05);
nclust = length(unique(part));
if nclust==3
mini_1 = min(ordered(find(part==1)));
mini_2 = min(ordered(find(part==2)));
mini_3 = min(ordered(find(part==3)));
if mini_1>0.9 & mini_2>0.9
part(find(part==2)) = 1;
part(find(part==3)) = 2;
elseif mini_1>0.9 & mini_3>0.9
part(find(part==3)) = 1;
elseif mini_2>0.9 & mini_3>0.9
% This is the one happening in practice, since the values are
% ordered, leading to mini_1 <= mini_2 <= mini_3
part(find(part==3)) = 2;
end
end

81
matlab/parallel/compare.m Normal file
View file

@ -0,0 +1,81 @@
function compare(varargin)
% COMPARE compares the results from multiple runs.
% input: is a group of result .mat files on the same data.
% Example: compare('e:\data\result1.mat','e:\data\result2.mat',...)
% or call it from the BAPS menu.
if nargin == 1
error('number of input arguments must be >=2');
end
if nargin == 0
out = uipickfiles('FilterSpec','*.mat',...
'Prompt','Select mixture results: be sure that the underlying data and models are consistent.');
if isnumeric(out)
return
end
nfiles = length(out);
filesin = out;
else
nfiles = nargin;
filesin = varargin;
end
display('---------------------------------------------------');
fprintf(1,'Comparing results ...\n');
if nfiles == 1
disp('*** ERROR: Too few files.');
return
end
for i = 1:nfiles
struct_array = load(filesin{i});
if isfield(struct_array,'c') %Matlab versio
c = struct_array.c;
clear struct_array;
if ~isfield(c,'PARTITION') || ~isfield(c,'rowsFromInd') || strcmp(c.mixtureType,'admix')
fprintf(1,'*** ERROR: Incorrect mixture result in file %d\n',i );
return
end
elseif isfield(struct_array,'PARTITION') %Mideva versio
c = struct_array;
if ~isfield(c,'rowsFromInd')
fprintf(1,'*** ERROR: Incorrect mixture result in file %d\n',i );
return
end
else
fprintf(1,'*** ERROR: Incorrect mixture result in file %d\n',i );
return;
end
try
partitionMat(i,:) = sort_partition(c.PARTITION);
catch
error('*** ERROR: inconsistent results.');
end
mixtureType{i} = c.mixtureType;
logml(i) = c.logml;
clear c;
end
len_mixture = length(mixtureType{1});
for i = 2:nfiles
if len_mixture ~= length(mixtureType{i});
error('*** ERROR: inconsistent mixture types.');
end
end
% Find the best partition
best = logical(logml == max(logml));
[uniquepartition, ind1, ind2] = unique(partitionMat(best,:), 'rows');
fprintf(1,'Best partition was found at ''%s''\n',filesin{best});

142
matlab/parallel/compare_admix.m Normal file
View file

@ -0,0 +1,142 @@
function compare_admix(varargin)
% COMPARE compares the results from multiple runs for admixture results
% input: is a group of result .mat files on the same data.
% Example: compare('e:\data\result1.mat','e:\data\result2.mat',...)
% or call it from the BAPS menu.
if nargin == 1
error('number of input arguments must be >=2');
end
if nargin == 0
out = uipickfiles('FilterSpec','*.mat',...
'Prompt','Select admixture results: be sure that the underlying data and parameters are consistent.');
if isnumeric(out)
return
end
nfiles = length(out);
filesin = out;
else
nfiles = nargin;
filesin = varargin;
end
display('---------------------------------------------------');
fprintf(1,'Comparing results ...\n');
minsize = zeros(nfiles,1);
iters = zeros(nfiles,1);
refInds = zeros(nfiles,1);
refIters = zeros(nfiles,1);
prop = cell(nfiles,1);
clusters = cell(nfiles,1);
adjprior = [];
if nfiles == 1
disp('*** ERROR: Too few files.');
return
end
% read admixture files
for i = 1:nfiles
struct_array = load(filesin{i});
if isfield(struct_array,'c') %Matlab versio
c = struct_array.c;
clear struct_array;
if ~isfield(c,'PARTITION') || ~isfield(c,'rowsFromInd') ...
|| ~isfield(c,'proportionsIt')
fprintf(1,'*** ERROR: Incorrect admixture result in file %d\n',i );
return
end
elseif isfield(struct_array,'PARTITION') %Mideva versio
c = struct_array;
if ~isfield(c,'rowsFromInd')
fprintf(1,'*** ERROR: Incorrect admixture result in file %d\n',i );
return
end
else
fprintf(1,'*** ERROR: Incorrect admixture result in file %d\n',i );
return;
end
prop{i} = c.proportionsIt;
pvalue(:,i) = c.pvalue;
clusters{i} = c.clusters;
popnames = c.popnames;
% parameters
minsize(i) = c.minsize;
iters(i) = c.iters;
refInds(i) = c.refInds;
refIters(i) = c.refIters;
if i==1
adjprior = c.adjprior;
else
if ~isequal(adjprior,c.adjprior)
disp('*** ERROR: incosistent admixture results.');
return
end
end
clear c;
end
if length(unique(minsize))~=1 || length(unique(iters))~=1 ...
|| length(unique(refInds))~=1 || length(unique(refIters))~=1
disp('*** ERROR: inconsistent admixture parameters.');
return
end
% now combine the results
prop_combine = prop{1};
[ninds npops] = size(prop_combine);
[pvalue_combine,index] = min(pvalue,[],2);
for i = 1:ninds
prop_combine(i,:) = prop{index(i)}(i,:);
end
% display the results
tulostaAdmixtureTiedot(prop_combine, pvalue_combine, minsize(1), iters(1));
viewPartition(prop_combine, popnames);
% save the results
talle = questdlg(['Do you want to save the combined admixture results?'], ...
'Save results?','Yes','No','Yes');
if isequal(talle,'Yes')
%waitALittle;
[filename, pathname] = uiputfile('*.mat','Save results as');
if (filename == 0) & (pathname == 0)
% Cancel was pressed
return
else % copy 'baps4_output.baps' into the text file with the same name.
if exist('baps4_output.baps','file')
copyfile('baps4_output.baps',[pathname filename '.txt'])
delete('baps4_output.baps')
end
end
struct_array = load(filesin{1});
c = struct_array.c;
clear struct_array;
c.proportionsIt = prop_combine;
c.pvalue = pvalue_combine; % Added by Jing
fprintf(1, 'Saving the results...');
% save([pathname filename], 'c');
save([pathname filename], 'c','-v7.3'); % added by Lu Cheng, 08.06.2012
fprintf(1,'finished.\');
end

3
matlab/parallel/dispLine.m Normal file
View file

@ -0,0 +1,3 @@
%------------------------------------------------------
function dispLine;
disp('---------------------------------------------------');

1622
matlab/parallel/greedyPopMix_parallel.m Normal file
View file

File diff suppressed because it is too large Load diff

1674
matlab/parallel/independent_parallel.m Normal file
View file

File diff suppressed because it is too large Load diff

35
matlab/parallel/initPopNames.m Normal file
View file

@ -0,0 +1,35 @@
function popnames = initPopNames(nameFile, indexFile)
%Palauttaa tyhjän, mikäli nimitiedosto ja indeksitiedosto
% eivät olleet yhtä pitkiä.
popnames = [];
indices = load(indexFile);
fid = fopen(nameFile);
if fid == -1
%File didn't exist
msgbox('Loading of the population names was unsuccessful', ...
'Error', 'error');
return;
end;
line = fgetl(fid);
counter = 1;
while (line ~= -1) && ~isempty(line)
names{counter} = line;
line = fgetl(fid);
counter = counter + 1;
end;
fclose(fid);
if length(names) ~= length(indices)
disp('The number of population names must be equal to the number of ');
disp('entries in the file specifying indices of the first individuals of ');
disp('each population.');
return;
end
popnames = cell(length(names), 2);
for i = 1:length(names)
popnames{i,1} = names(i);
popnames{i,2} = indices(i);
end

2138
matlab/parallel/linkageMixture_speed.m Normal file
View file

File diff suppressed because it is too large Load diff

1030
matlab/parallel/linkage_admix_parallel.m Normal file
View file

File diff suppressed because it is too large Load diff

1633
matlab/parallel/linkage_parallel.m Normal file
View file

File diff suppressed because it is too large Load diff

297
matlab/parallel/parallel.m Normal file
View file

@ -0,0 +1,297 @@
function parallel(varargin)
% PARALLEL the main function of doing parallel classification/admixture.
% input: order of input is arbitrary. The first option is the default.
% 'datafile' - the full path of the data.
% 'mixturetype', ['mix';'codon_mix';'linear_mix';'spatial';'ad_mix'];
% - the classification/admixture model.
% 'initialk', a row vector of positive integers;
% - the initial number of clusters.
% 'fixedk',['no';'yes'];
% - whether the number of clusters is fixed during the computation.
% 'outputmat' - the full path of the output .mat file.
% 'datatype', ['numeric';'sequence';'matlab';'excel';'genepop']
% - the data format;
% 'namefile' - the full path of the population name file.
% 'indexfile' - the full path of the index file.
% 'linkagemap' - the full path of the linkage map, needed only for the
% unpreprocessed data under the linkage model.
% 'coordinatefile' - needed with the spatial model.
% 'groups', ['no';'yes'] - clustering of groups instead of individuals.
% 'groupname' - the full path of the group name file.
% Examples:
% - Linkage model:
% parallel('datafile','e:\baps4\baps_source\data\bpseudomallei.xls',...
% 'mixturetype','codon_mix','initialk','[10 15]','fixedk','no',...
% 'outputmat','E:\test_link.mat','datatype','excel')
% - Independent model:
% parallel('datafile','e:\baps4\baps_source\data\baps_data.txt',...
% 'mixturetype','mix','initialk','[10:15]','fixedk','no',...
% 'outputmat','e:\test_ind.mat','datatype','numeric');
% - Spatial model:
% parallel('datafile','e:\baps4\baps_source\data\wolverines_spatial_preprocessed.mat',...
% 'mixturetype','spatial','initialk', '[10 11]', 'fixedk','no',...
% 'outputmat','e:\test_spatial.mat','datatype','matlab');
%
%
% - Admixture model:
% parallel('datafile','e:\baps5\data\data1_mixture.mat', ...
% 'mixturetype','ad_mix', ...
% 'clusters','[1 3 5]',...
% 'iters','2',...
% 'refinds','3',...
% 'refiters','4',...
% 'outputmat','e:\baps5\data\data1_admixture_parallel.mat');
% A group of result files can be later compared by using compare.m
% function.
%-------------------------------------------------------------------------------
%- Set up options and default parameters
%-------------------------------------------------------------------------------
msgInvalidPair = '***ERROR: Bad value for argument: ''%s''';
% default options
options = struct('dataFile', '',...
'dataType', 'numeric',...
'mixtureType', 'mix',...
'initialK', 1, ...
'fixedK', 'no', ...
'nameFile', '', ...
'indexFile', '', ...
'outputMat', '', ...
'linkageMap','', ...
'coordinateFile', '', ...
'groups','no', ...
'groupname', '', ...
'clusters', '', ...
'minSize', '', ...
'iters', '', ...
'refInds', '', ...
'refIters', '' ...
);
if nargin == 1 && isstruct(varargin{1})
paramlist = [ fieldnames(varargin{1}) ...
struct2cell(varargin{1}) ]';
paramlist = { paramlist{:} };
else
if mod(nargin,2)
error('Invalid parameter/value pair arguments.');
end
paramlist = varargin;
end
optionsnames = lower(fieldnames(options));
for i=1:2:length(paramlist)
pname = paramlist{i};
pvalue = paramlist{i+1};
ind = strmatch(lower(pname),optionsnames);
if isempty(ind)
error(['Invalid parameter: ''' pname '''.']);
elseif length(ind) > 1
error(['Ambiguous parameter: ''' pname '''.']);
end
switch(optionsnames{ind})
case 'datafile'
if ischar(pvalue)
options.dataFile = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
% if ~isempty(findstr(pvalue , '.txt'))
% options.dataType = 'text';
% elseif ~isempty(findstr(pvalue, '.mat'))
% options.dataType = 'matlab';
% elseif ~isempty(findstr(pvalue, '.xls'))
% options.dataType = 'excel';
% else
% error('*** ERROR: unrecognized data format');
% end
case 'mixturetype'
if ischar(pvalue)
if ~strmatch(pvalue, strvcat('mix','ad_mix','linear_mix','codon_mix','spatical'),'exact')
error('*** ERROR: unrecoganized model type');
end
if isempty(pvalue),
options.mixtureType = '.';
else
options.mixtureType = pvalue;
end
else
error(sprintf(msgInvalidPair,pname));
end
case 'initialk'
pvalue = str2num(pvalue);
if isnumeric(pvalue)
if isempty(pvalue),
options.initialK = 0;
else
options.initialK = pvalue;
end
else
error(sprintf(msgInvalidPair,pname));
end
case 'fixedk'
if ischar(pvalue)
if isempty(pvalue),
options.fixedK = 'no';
else
options.fixedK = pvalue;
end
else
error(sprintf(msgInvalidPair,pname));
end
case 'namefile'
if ischar(pvalue)
options.nameFile = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
case 'indexfile'
if ischar(pvalue)
options.indexFile = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
case 'outputmat'
if ischar(pvalue)
options.outputMat = pvalue;
directoryName = fileparts(pvalue);
if ~exist(directoryName)
fprintf(1,'*** ERROR: Output directory ''%s'' does not exist.\n', directoryName);
return
end
else
error(sprintf(msgInvalidPair,pname));
end
case 'datatype'
if ischar(pvalue)
options.dataType = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
case 'linkagemap'
if ischar(pvalue)
options.linkageMap = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
case 'coordinatefile'
if ischar(pvalue)
options.coordinateFile = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
case 'groups'
if ischar(pvalue)
options.groups = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
case 'groupname'
if ischar(pvalue)
options.groupname = pvalue;
else
error(sprintf(msgInvalidPair,pname));
end
% the options below are for admixture analysis
case 'clusters'
if ischar(pvalue)
options.clusters = str2num(pvalue);
else
error(sprintf(msgInvalidPair,pname));
end
case 'minsize'
if ischar(pvalue)
options.minSize = str2num(pvalue);
else
error(sprintf(msgInvalidPair,pname));
end
case 'iters'
if ischar(pvalue)
options.iters = str2num(pvalue);
else
error(sprintf(msgInvalidPair,pname));
end
case 'refinds'
if ischar(pvalue)
options.refInds = str2num(pvalue);
else
error(sprintf(msgInvalidPair,pname));
end
case 'refiters'
if ischar(pvalue)
options.refIters = str2num(pvalue);
else
error(sprintf(msgInvalidPair,pname));
end
otherwise
error(['Invalid parameter: ''' pname '''.']);
end
end
% The subfunction to check syntax
if ~checkSyntax(options)
return
end
switch options.mixtureType
case 'mix'
if isequal(options.groups,'yes')
greedyPopMix_parallel(options);
else
independent_parallel(options);
end
case 'linear_mix'
linkage_parallel(options);
case 'codon_mix'
linkage_parallel(options);
case 'spatial'
if isequal(options.groups, 'yes')
spatialPopMixture_parallel(options);
else
spatial_parallel(options);
end
case 'ad_mix'
admix_parallel(options);
end
% -------------------------------------------------------------------------
% Subfunctions
% -------------------------------------------------------------------------
function isOK = checkSyntax(options)
isOK = 1;
if strcmp(options.fixedK, 'yes') && length(options.initialK)>1
display('*** ERROR: conflicting in options fixedk and initialk.');
isOK = 0;
end
if strcmp(options.mixtureType, 'mix')
if strcmp(options.dataType, 'excel') || strcmp(options.dataType,'sequence')
display('*** ERROR: unknown datatype for the independence module.');
isOK = 0;
end
end
% check the admixture parameters
admix_str = {options.clusters, options.minSize, options.iters, ...
options.refInds, options.refIters};
pt = cellfun('isempty', admix_str);
if all(pt)
if ~strcmp(options.mixtureType, 'ad_mix')
isOK = 1;
else
display('*** ERROR: problematic mixture type.');
isOK = 0;
end
end
if any(pt) && strcmp(options.mixtureType, 'ad_mix')
display('*** ERROR: incomplete admixture parameters.');
isOK = 0;
else
isOK = 1;
end

18
matlab/parallel/proportion2str.m Normal file
View file

@ -0,0 +1,18 @@
function str = proportion2str(prob)
%prob belongs to [0.00, 0.01, ... ,1].
%str is a 4-mark presentation of proportion.
if abs(prob)<1e-3
str = '0.00';
elseif abs(prob-1) < 1e-3;
str = '1.00';
else
prob = round(100*prob);
if prob<10
str = ['0.0' num2str(prob)];
else
str = ['0.' num2str(prob)];
end;
end;
%-------------------------------------------------------

66
matlab/parallel/readScript.m Normal file
View file

@ -0,0 +1,66 @@
function readScript(filename)
% READSCRIPT read the script file and output the parameters
% this function does not perform syntax checking.
% Example:
% readScript('script.txt')
% read the script
ind = readfile(filename);
if isempty(ind)
return
end
nLines = size(ind,1);
% extract command information
optionStr = [];
for k = 1:nLines
[cmdName, paraStr] = extract(ind(k,:));
optionStr = [optionStr cmdName ',' paraStr ','];
end
optionStr = optionStr(1:end-1); % remove the last coma
% call function parallel
eval(['parallel(' optionStr ')'])
% -------------------------------------------------------------------------
% Subfunctions
% -------------------------------------------------------------------------
function [cmdName, paraStr] = extract(commandline)
% function to extract the command name and the parameter string
[cmdName, remainStr] = strtok(commandline,'(');
boundary = regexp(remainStr,'''');
if isempty(boundary) % if paraStr does not contain quotation marks
% use parenthesis as boundaries
startPt = regexp(remainStr,'(') + 1;
endPt = regexp(remainStr,')') - 1;
else
startPt = boundary(1) + 1;
endPt = boundary(2) - 1;
end
paraStr = remainStr(startPt: endPt);
cmdName = strcat('''',cmdName,'''');
paraStr = strcat('''',paraStr,'''');
% -------------------------------------------------------------------------
function T = readfile(filename);
f = fopen(filename,'r');
if f == -1
% error(filename);
display('*** ERROR: invalid script name.');
T = [];
return
end
i = 1;
while 1
clear line;
line = fgetl(f);
if ~isstr(line), break, end
n = length(line);
T(i,1:n) = line(1:n);
i = i+1;
end
fclose(f);

6
matlab/parallel/sc.txt Normal file
View file

@ -0,0 +1,6 @@
datafile('r:\baps5\data\simple_data.txt');
mixturetype('mix')
outputmat('r:\baps5\data\simple_mix.mat')
initialk(2 3)
fixedk('no')
datatype('numeric')

8
matlab/parallel/sc2.txt Normal file
View file

@ -0,0 +1,8 @@
datafile('e:\baps5\data\ssuis_mixture_results_codon.mat')
mixturetype('ad_mix')
outputmat('e:\baps5\data\ssuis_small_script_2.mat')
clusters(3:4)
minsize(5)
iters(2)
refinds(3)
refiters(4)

17
matlab/parallel/sort_partition.m Normal file
View file

@ -0,0 +1,17 @@
function srt_partition = sort_partition(partition)
% SORT_PARTITION sorts a given partition (row vector) into the canonical order, where every
% new class has the smallest possible index.
% input:
% partition is a column vector.
% output:
% srt_partition is a row vector.
n_classes=max(partition);
srt_partition=zeros(1,n_classes);
for i=1:n_classes
nonz=find(partition);
here=find(partition==partition(nonz(1)));
srt_partition(here)=i;
partition(here)=0;
end

2275
matlab/parallel/spatialPopMixture_parallel.m Normal file
View file

File diff suppressed because it is too large Load diff

2175
matlab/parallel/spatial_parallel.m Normal file
View file

File diff suppressed because it is too large Load diff

63
matlab/parallel/tulostaAdmixtureTiedot.m Normal file
View file

@ -0,0 +1,63 @@
%--------------------------------------------------------------------------
function tulostaAdmixtureTiedot(proportions, uskottavuus, alaRaja, niter)
h0 = findobj('Tag','filename1_text');
inputf = get(h0,'String');
h0 = findobj('Tag','filename2_text');
outf = get(h0,'String'); clear h0;
if length(outf)>0
fid = fopen(outf,'a');
else
fid = -1;
diary('baps4_output.baps'); % save in text anyway.
end
ninds = length(uskottavuus);
npops = size(proportions,2);
disp(' ');
dispLine;
disp('RESULTS OF ADMIXTURE ANALYSIS BASED');
disp('ON MIXTURE CLUSTERING OF INDIVIDUALS');
disp(['Data file: ' inputf]);
disp(['Number of individuals: ' num2str(ninds)]);
disp(['Results based on ' num2str(niter) ' simulations from posterior allele frequencies.']);
disp(' ');
if fid ~= -1
fprintf(fid, '\n');
fprintf(fid,'%s \n', ['--------------------------------------------']); fprintf(fid, '\n');
fprintf(fid,'%s \n', ['RESULTS OF ADMIXTURE ANALYSIS BASED']); fprintf(fid, '\n');
fprintf(fid,'%s \n', ['ON MIXTURE CLUSTERING OF INDIVIDUALS']); fprintf(fid, '\n');
fprintf(fid,'%s \n', ['Data file: ' inputf]); fprintf(fid, '\n');
fprintf(fid,'%s \n', ['Number of individuals: ' num2str(ninds)]); fprintf(fid, '\n');
fprintf(fid,'%s \n', ['Results based on ' num2str(niter) ' simulations from posterior allele frequencies.']); fprintf(fid, '\n');
fprintf(fid, '\n');
end
ekaRivi = blanks(6);
for pop = 1:npops
ekaRivi = [ekaRivi blanks(3-floor(log10(pop))) num2str(pop) blanks(2)];
end
ekaRivi = [ekaRivi blanks(1) 'p']; % Added on 29.08.06
disp(ekaRivi);
for ind = 1:ninds
rivi = [num2str(ind) ':' blanks(4-floor(log10(ind)))];
if any(proportions(ind,:)>0)
for pop = 1:npops-1
rivi = [rivi proportion2str(proportions(ind,pop)) blanks(2)];
end
rivi = [rivi proportion2str(proportions(ind,npops)) ': '];
rivi = [rivi ownNum2Str(uskottavuus(ind))];
end
disp(rivi);
if fid ~= -1
fprintf(fid,'%s \n',[rivi]); fprintf(fid,'\n');
end
end
if fid ~= -1
fclose(fid);
else
diary off
end

803
matlab/parallel/uipickfiles.m Normal file
View file

@ -0,0 +1,803 @@
function out = uipickfiles(varargin)
%uipickfiles: GUI program to select file(s) and/or directories.
%
% Syntax:
% files = uipickfiles('PropertyName',PropertyValue,...)
%
% The current directory can be changed by operating in the file navigator:
% double-clicking on a directory in the list to move further down the tree,
% using the popup menu to move up the tree, typing a path in the box to
% move to any directory or right-clicking on the path box to revisit a
% previously-listed directory.
%
% Files can be added to the list by double-clicking or selecting files
% (non-contiguous selections are possible with the control key) and
% pressing the Add button. Files in the list can be removed or re-ordered.
% When finished, a press of the Done button will return the full paths to
% the selected files in a cell array, structure array or character array.
% If the Cancel button is pressed then zero is returned.
%
% The following optional property/value pairs can be specified as arguments
% to control the indicated behavior:
%
% Property Value
% ---------- ----------------------------------------------------------
% FilterSpec String to specify starting directory and/or file filter.
% Ex: 'C:\bin' will start up in that directory. '*.txt'
% will list only files ending in '.txt'. 'c:\bin\*.txt' will
% do both. Default is to start up in the current directory
% and list all files. Can be changed with the GUI.
%
% REFilter String containing a regular expression used to filter the
% file list. Ex: '\.m$|\.mat$' will list files ending in
% '.m' and '.mat'. Default is empty string. Can be used
% with FilterSpec and both filters are applied. Can be
% changed with the GUI.
%
% Prompt String containing a prompt appearing in the title bar of
% the figure. Default is 'Select files'.
%
% NumFiles Scalar or vector specifying number of files that must be
% selected. A scalar specifies an exact value; a two-element
% vector can be used to specify a range, [min max]. The
% function will not return unless the specified number of
% files have been chosen. Default is [] which accepts any
% number of files.
%
% Output String specifying the data type of the output: 'cell',
% 'struct' or 'char'. Specifying 'cell' produces a cell
% array of strings, the strings containing the full paths of
% the chosen files. 'Struct' returns a structure array like
% the result of the dir function except that the 'name' field
% contains a full path instead of just the file name. 'Char'
% returns a character array of the full paths. This is most
% useful when you have just one file and want it in a string
% instead of a cell array containing just one string. The
% default is 'cell'.
%
% All properties and values are case-insensitive and need only be
% unambiguous. For example,
%
% files = uipickfiles('num',1,'out','ch')
%
% is valid usage.
% Version: 1.0, 25 April 2006
% Author: Douglas M. Schwarz
% Email: dmschwarz=ieee*org, dmschwarz=urgrad*rochester*edu
% Real_email = regexprep(Email,{'=','*'},{'@','.'})
% Define properties and set default values.
prop.filterspec = '*';
prop.refilter = '';
prop.prompt = 'Select files';
prop.numfiles = [];
prop.output = 'cell';
% Process inputs and set prop fields.
properties = fieldnames(prop);
arg_index = 1;
while arg_index <= nargin
arg = varargin{arg_index};
if ischar(arg)
prop_index = find(strncmpi(arg,properties,length(arg)));
if length(prop_index) == 1
prop.(properties{prop_index}) = varargin{arg_index + 1};
else
error('Property ''%s'' does not exist or is ambiguous.',arg)
end
arg_index = arg_index + 2;
elseif isstruct(arg)
arg_fn = fieldnames(arg);
for i = 1:length(arg_fn)
prop_index = find(strncmpi(arg_fn{i},properties,...
length(arg_fn{i})));
if length(prop_index) == 1
prop.(properties{prop_index}) = arg.(arg_fn{i});
else
error('Property ''%s'' does not exist or is ambiguous.',...
arg_fn{i})
end
end
arg_index = arg_index + 1;
else
error(['Properties must be specified by property/value pairs',...
' or structures.'])
end
end
% Validate FilterSpec property.
if isempty(prop.filterspec)
prop.filterspec = '*';
end
if ~ischar(prop.filterspec)
error('FilterSpec property must contain a string.')
end
% Validate REFilter property.
if ~ischar(prop.refilter)
error('REFilter property must contain a string.')
end
% Validate Prompt property.
if ~ischar(prop.prompt)
error('Prompt property must contain a string.')
end
% Validate NumFiles property.
if numel(prop.numfiles) > 2 || any(prop.numfiles < 0)
error('NumFiles must be empty, a scalar or two-element vector.')
end
prop.numfiles = unique(prop.numfiles);
if isequal(prop.numfiles,1)
numstr = 'Select exactly 1 file.';
elseif length(prop.numfiles) == 1
numstr = sprintf('Select exactly %d files.',prop.numfiles);
else
numstr = sprintf('Select %d to %d files.',prop.numfiles);
end
% Validate Output property.
legal_outputs = {'cell','struct','char'};
out_idx = find(strncmpi(prop.output,legal_outputs,length(prop.output)));
if length(out_idx) == 1
prop.output = legal_outputs{out_idx};
else
error(['Value of ''Output'' property, ''%s'', is illegal or '...
'ambiguous.'],prop.output)
end
% Initialize file lists.
[current_dir,f,e] = fileparts(prop.filterspec);
filter = [f,e];
if isempty(current_dir)
current_dir = pwd;
end
if isempty(filter)
filter = '*';
end
re_filter = prop.refilter;
full_filter = fullfile(current_dir,filter);
path_cell = path2cell(current_dir);
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
% Initialize some data.
file_picks = {};
full_file_picks = {};
dir_picks = dir(' '); % Create empty directory structure.
show_full_path = false;
nodupes = true;
history = {current_dir};
% Create figure.
gray = get(0,'DefaultUIControlBackgroundColor');
fig = figure('Position',[0 0 740 445],...
'Color',gray,...
'WindowStyle','modal',...
'Resize','off',...
'NumberTitle','off',...
'Name',prop.prompt,...
'IntegerHandle','off',...
'CloseRequestFcn',@cancel,...
'CreateFcn',{@movegui,'center'});
% Create uicontrols.
uicontrol('Style','frame',...
'Position',[255 260 110 70])
uicontrol('Style','frame',...
'Position',[275 135 110 100])
navlist = uicontrol('Style','listbox',...
'Position',[10 10 250 320],...
'String',filenames,...
'Value',[],...
'BackgroundColor','w',...
'Callback',@clicknav,...
'Max',2);
pickslist = uicontrol('Style','listbox',...
'Position',[380 10 350 320],...
'String',{},...
'BackgroundColor','w',...
'Callback',@clickpicks,...
'Max',2);
openbut = uicontrol('Style','pushbutton',...
'Position',[270 300 80 20],...
'String','Open',...
'Enable','off',...
'Callback',@open);
arrow = [2 2 2 2 2 2 2 2 1 2 2 2;...
2 2 2 2 2 2 2 2 2 0 2 2;...
2 2 2 2 2 2 2 2 2 2 0 2;...
0 0 0 0 0 0 0 0 0 0 0 0;...
2 2 2 2 2 2 2 2 2 2 0 2;...
2 2 2 2 2 2 2 2 2 0 2 2;...
2 2 2 2 2 2 2 2 1 2 2 2];
arrow(arrow == 2) = NaN;
arrow_im = NaN*ones(16,76);
arrow_im(6:12,45:56) = arrow/2;
im = repmat(arrow_im,[1 1 3]);
addbut = uicontrol('Style','pushbutton',...
'Position',[270 270 80 20],...
'String','Add ',...
'Enable','off',...
'CData',im,...
'Callback',@add);
removebut = uicontrol('Style','pushbutton',...
'Position',[290 205 80 20],...
'String','Remove',...
'Enable','off',...
'Callback',@remove);
moveupbut = uicontrol('Style','pushbutton',...
'Position',[290 175 80 20],...
'String','Move Up',...
'Enable','off',...
'Callback',@moveup);
movedownbut = uicontrol('Style','pushbutton',...
'Position',[290 145 80 20],...
'String','Move Down',...
'Enable','off',...
'Callback',@movedown);
uicontrol('Position',[10 380 250 16],...
'Style','text',...
'String','Current Directory',...
'HorizontalAlignment','center')
dir_popup = uicontrol('Style','popupmenu',...
'Position',[10 335 250 20],...
'BackgroundColor','w',...
'String',path_cell(end:-1:1),...
'Value',1,...
'Callback',@dirpopup);
hist_cm = uicontextmenu;
pathbox = uicontrol('Position',[10 360 250 20],...
'Style','edit',...
'BackgroundColor','w',...
'String',current_dir,...
'HorizontalAlignment','left',...
'Callback',@change_path,...
'UIContextMenu',hist_cm);
hist_menus = [];
hist_cb = @history_cb;
hist_menus = make_history_cm(hist_cb,hist_cm,hist_menus,history);
uicontrol('Position',[10 425 80 16],...
'Style','text',...
'String','File Filter',...
'HorizontalAlignment','left')
uicontrol('Position',[100 425 160 16],...
'Style','text',...
'String','Reg. Exp. Filter',...
'HorizontalAlignment','left')
showallfiles = uicontrol('Position',[270 405 100 20],...
'Style','checkbox',...
'String','Show All Files',...
'Value',0,...
'HorizontalAlignment','left',...
'Callback',@togglefilter);
filter_ed = uicontrol('Position',[10 405 80 20],...
'Style','edit',...
'BackgroundColor','w',...
'String',filter,...
'HorizontalAlignment','left',...
'Callback',@setfilspec);
refilter_ed = uicontrol('Position',[100 405 160 20],...
'Style','edit',...
'BackgroundColor','w',...
'String',re_filter,...
'HorizontalAlignment','left',...
'Callback',@setrefilter);
viewfullpath = uicontrol('Style','checkbox',...
'Position',[380 335 230 20],...
'String','Show full paths',...
'Value',show_full_path,...
'HorizontalAlignment','left',...
'Callback',@showfullpath);
remove_dupes = uicontrol('Style','checkbox',...
'Position',[380 360 230 20],...
'String','Remove duplicates (as per full path)',...
'Value',nodupes,...
'HorizontalAlignment','left',...
'Callback',@removedupes);
uicontrol('Position',[380 405 350 20],...
'Style','text',...
'String','Selected Files',...
'HorizontalAlignment','center')
uicontrol('Position',[280 80 80 30],'String','Done',...
'Callback',@done);
uicontrol('Position',[280 30 80 30],'String','Cancel',...
'Callback',@cancel);
if ~isempty(prop.numfiles)
uicontrol('Position',[380 385 350 16],...
'Style','text',...
'String',numstr,...
'ForegroundColor','r',...
'HorizontalAlignment','center')
end
set(fig,'HandleVisibility','off')
uiwait(fig)
% Compute desired output.
switch prop.output
case 'cell'
out = full_file_picks;
case 'struct'
out = dir_picks(:);
case 'char'
out = char(full_file_picks);
case 'cancel'
out = 0;
end
% -------------------- Callback functions --------------------
function add(varargin)
values = get(navlist,'Value');
for i = 1:length(values)
dir_pick = fdir(values(i));
pick = dir_pick.name;
pick_full = fullfile(current_dir,pick);
dir_pick.name = pick_full;
if ~nodupes || ~any(strcmp(full_file_picks,pick_full))
file_picks{end + 1} = pick;
full_file_picks{end + 1} = pick_full;
dir_picks(end + 1) = dir_pick;
end
end
if show_full_path
set(pickslist,'String',full_file_picks,'Value',[]);
else
set(pickslist,'String',file_picks,'Value',[]);
end
set([removebut,moveupbut,movedownbut],'Enable','off');
end
function remove(varargin)
values = get(pickslist,'Value');
file_picks(values) = [];
full_file_picks(values) = [];
dir_picks(values) = [];
top = get(pickslist,'ListboxTop');
num_above_top = sum(values < top);
top = top - num_above_top;
num_picks = length(file_picks);
new_value = min(min(values) - num_above_top,num_picks);
if num_picks == 0
new_value = [];
set([removebut,moveupbut,movedownbut],'Enable','off')
end
if show_full_path
set(pickslist,'String',full_file_picks,'Value',new_value,...
'ListboxTop',top)
else
set(pickslist,'String',file_picks,'Value',new_value,...
'ListboxTop',top)
end
end
function open(varargin)
values = get(navlist,'Value');
if fdir(values).isdir
if strcmp(fdir(values).name,'.')
return
elseif strcmp(fdir(values).name,'..')
set(dir_popup,'Value',min(2,length(path_cell)))
dirpopup();
return
end
current_dir = fullfile(current_dir,fdir(values).name);
history{end+1} = current_dir;
history = unique(history);
hist_menus = make_history_cm(hist_cb,hist_cm,hist_menus,...
history);
full_filter = fullfile(current_dir,filter);
path_cell = path2cell(current_dir);
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(dir_popup,'String',path_cell(end:-1:1),'Value',1)
set(pathbox,'String',current_dir)
set(navlist,'ListboxTop',1,'Value',[],'String',filenames)
set(addbut,'Enable','off')
set(openbut,'Enable','off')
end
end
function clicknav(varargin)
value = get(navlist,'Value');
nval = length(value);
dbl_click_fcn = @add;
switch nval
case 0
set([addbut,openbut],'Enable','off')
case 1
set(addbut,'Enable','on');
if fdir(value).isdir
set(openbut,'Enable','on')
dbl_click_fcn = @open;
else
set(openbut,'Enable','off')
end
otherwise
set(addbut,'Enable','on')
set(openbut,'Enable','off')
end
if strcmp(get(fig,'SelectionType'),'open')
dbl_click_fcn();
end
end
function clickpicks(varargin)
value = get(pickslist,'Value');
if isempty(value)
set([removebut,moveupbut,movedownbut],'Enable','off')
else
set(removebut,'Enable','on')
if min(value) == 1
set(moveupbut,'Enable','off')
else
set(moveupbut,'Enable','on')
end
if max(value) == length(file_picks)
set(movedownbut,'Enable','off')
else
set(movedownbut,'Enable','on')
end
end
if strcmp(get(fig,'SelectionType'),'open')
remove();
end
end
function dirpopup(varargin)
value = get(dir_popup,'Value');
len = length(path_cell);
path_cell = path_cell(1:end-value+1);
if ispc && value == len
current_dir = '';
full_filter = filter;
fdir = struct('name',getdrives,'date',datestr(now),...
'bytes',0,'isdir',1);
else
current_dir = cell2path(path_cell);
history{end+1} = current_dir;
history = unique(history);
hist_menus = make_history_cm(hist_cb,hist_cm,hist_menus,...
history);
full_filter = fullfile(current_dir,filter);
fdir = filtered_dir(full_filter,re_filter);
end
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(dir_popup,'String',path_cell(end:-1:1),'Value',1)
set(pathbox,'String',current_dir)
set(navlist,'String',filenames,'Value',[])
set(addbut,'Enable','off')
end
function change_path(varargin)
proposed_path = get(pathbox,'String');
% Process any directories named '..'.
proposed_path_cell = path2cell(proposed_path);
ddots = strcmp(proposed_path_cell,'..');
ddots(find(ddots) - 1) = true;
proposed_path_cell(ddots) = [];
proposed_path = cell2path(proposed_path_cell);
% Check for existance of directory.
if ~exist(proposed_path,'dir')
uiwait(errordlg(['Directory "',proposed_path,...
'" does not exist.'],'','modal'))
return
end
current_dir = proposed_path;
history{end+1} = current_dir;
history = unique(history);
hist_menus = make_history_cm(hist_cb,hist_cm,hist_menus,history);
full_filter = fullfile(current_dir,filter);
path_cell = path2cell(current_dir);
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(dir_popup,'String',path_cell(end:-1:1),'Value',1)
set(pathbox,'String',current_dir)
set(navlist,'String',filenames,'Value',[])
set(addbut,'Enable','off')
set(openbut,'Enable','off')
end
function showfullpath(varargin)
show_full_path = get(viewfullpath,'Value');
if show_full_path
set(pickslist,'String',full_file_picks)
else
set(pickslist,'String',file_picks)
end
end
function removedupes(varargin)
nodupes = get(remove_dupes,'Value');
if nodupes
num_picks = length(full_file_picks);
[unused,rev_order] = unique(full_file_picks(end:-1:1));
order = sort(num_picks + 1 - rev_order);
full_file_picks = full_file_picks(order);
file_picks = file_picks(order);
if show_full_path
set(pickslist,'String',full_file_picks,'Value',[])
else
set(pickslist,'String',file_picks,'Value',[])
end
set([removebut,moveupbut,movedownbut],'Enable','off')
end
end
function moveup(varargin)
value = get(pickslist,'Value');
set(removebut,'Enable','on')
n = length(file_picks);
omega = 1:n;
index = zeros(1,n);
index(value - 1) = omega(value);
index(setdiff(omega,value - 1)) = omega(setdiff(omega,value));
file_picks = file_picks(index);
full_file_picks = full_file_picks(index);
value = value - 1;
if show_full_path
set(pickslist,'String',full_file_picks,'Value',value)
else
set(pickslist,'String',file_picks,'Value',value)
end
if min(value) == 1
set(moveupbut,'Enable','off')
end
set(movedownbut,'Enable','on')
end
function movedown(varargin)
value = get(pickslist,'Value');
set(removebut,'Enable','on')
n = length(file_picks);
omega = 1:n;
index = zeros(1,n);
index(value + 1) = omega(value);
index(setdiff(omega,value + 1)) = omega(setdiff(omega,value));
file_picks = file_picks(index);
full_file_picks = full_file_picks(index);
value = value + 1;
if show_full_path
set(pickslist,'String',full_file_picks,'Value',value)
else
set(pickslist,'String',file_picks,'Value',value)
end
if max(value) == n
set(movedownbut,'Enable','off')
end
set(moveupbut,'Enable','on')
end
function togglefilter(varargin)
value = get(showallfiles,'Value');
if value
filter = '*';
re_filter = '';
set([filter_ed,refilter_ed],'Enable','off')
else
filter = get(filter_ed,'String');
re_filter = get(refilter_ed,'String');
set([filter_ed,refilter_ed],'Enable','on')
end
full_filter = fullfile(current_dir,filter);
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(navlist,'String',filenames,'Value',[])
set(addbut,'Enable','off')
end
function setfilspec(varargin)
filter = get(filter_ed,'String');
if isempty(filter)
filter = '*';
set(filter_ed,'String',filter)
end
% Process file spec if a subdirectory was included.
[p,f,e] = fileparts(filter);
if ~isempty(p)
newpath = fullfile(current_dir,p,'');
set(pathbox,'String',newpath)
filter = [f,e];
if isempty(filter)
filter = '*';
end
set(filter_ed,'String',filter)
change_path();
end
full_filter = fullfile(current_dir,filter);
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(navlist,'String',filenames,'Value',[])
set(addbut,'Enable','off')
end
function setrefilter(varargin)
re_filter = get(refilter_ed,'String');
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(navlist,'String',filenames,'Value',[])
set(addbut,'Enable','off')
end
function done(varargin)
% Optional shortcut: click on a file and press 'Done'.
% if isempty(full_file_picks) && strcmp(get(addbut,'Enable'),'on')
% add();
% end
numfiles = length(full_file_picks);
if ~isempty(prop.numfiles)
if numfiles < prop.numfiles(1)
msg = {'Too few files selected.',numstr};
uiwait(errordlg(msg,'','modal'))
return
elseif numfiles > prop.numfiles(end)
msg = {'Too many files selected.',numstr};
uiwait(errordlg(msg,'','modal'))
return
end
end
delete(fig)
end
function cancel(varargin)
prop.output = 'cancel';
delete(fig)
end
function history_cb(varargin)
current_dir = history{varargin{3}};
full_filter = fullfile(current_dir,filter);
path_cell = path2cell(current_dir);
fdir = filtered_dir(full_filter,re_filter);
filenames = {fdir.name}';
filenames = annotate_file_names(filenames,fdir);
set(dir_popup,'String',path_cell(end:-1:1),'Value',1)
set(pathbox,'String',current_dir)
set(navlist,'ListboxTop',1,'Value',[],'String',filenames)
set(addbut,'Enable','off')
set(openbut,'Enable','off')
end
end
% -------------------- Subfunctions --------------------
function c = path2cell(p)
% Turns a path string into a cell array of path elements.
c = strread(p,'%s','delimiter','\\/');
if ispc
c = [{'My Computer'};c];
else
c = [{filesep};c(2:end)];
end
end
function p = cell2path(c)
% Turns a cell array of path elements into a path string.
if ispc
p = fullfile(c{2:end},'');
else
p = fullfile(c{:},'');
end
end
function d = filtered_dir(full_filter,re_filter)
% Like dir, but applies filters and sorting.
p = fileparts(full_filter);
if isempty(p) && full_filter(1) == '/'
p = '/';
end
if exist(full_filter,'dir')
c = cell(0,1);
dfiles = struct('name',c,'date',c,'bytes',c,'isdir',c);
else
dfiles = dir(full_filter);
end
if ~isempty(dfiles)
dfiles([dfiles.isdir]) = [];
end
ddir = dir(p);
ddir = ddir([ddir.isdir]);
% Additional regular expression filter.
if nargin > 1 && ~isempty(re_filter)
if ispc
no_match = cellfun('isempty',regexpi({dfiles.name},re_filter));
else
no_match = cellfun('isempty',regexp({dfiles.name},re_filter));
end
dfiles(no_match) = [];
end
% Set navigator style:
% 1 => mix file and directory names
% 2 => means list all files before all directories
% 3 => means list all directories before all files
% 4 => same as 2 except put . and .. directories first
if isunix
style = 4;
else
style = 4;
end
switch style
case 1
d = [dfiles;ddir];
[unused,index] = sort({d.name});
d = d(index);
case 2
[unused,index1] = sort({dfiles.name});
[unused,index2] = sort({ddir.name});
d = [dfiles(index1);ddir(index2)];
case 3
[unused,index1] = sort({dfiles.name});
[unused,index2] = sort({ddir.name});
d = [ddir(index2);dfiles(index1)];
case 4
[unused,index1] = sort({dfiles.name});
dot1 = find(strcmp({ddir.name},'.'));
dot2 = find(strcmp({ddir.name},'..'));
ddot1 = ddir(dot1);
ddot2 = ddir(dot2);
ddir([dot1,dot2]) = [];
[unused,index2] = sort({ddir.name});
d = [ddot1;ddot2;dfiles(index1);ddir(index2)];
end
end
function drives = getdrives
% Returns a cell array of drive names on Windows.
letters = char('A':'Z');
num_letters = length(letters);
drives = cell(1,num_letters);
for i = 1:num_letters
if exist([letters(i),':\'],'dir');
drives{i} = [letters(i),':'];
end
end
drives(cellfun('isempty',drives)) = [];
end
function filenames = annotate_file_names(filenames,dir_listing)
% Adds a trailing filesep character to directory names.
fs = filesep;
for i = 1:length(filenames)
if dir_listing(i).isdir
filenames{i} = [filenames{i},fs];
end
end
end
function hist_menus = make_history_cm(cb,hist_cm,hist_menus,history)
% Make context menu for history.
if ~isempty(hist_menus)
delete(hist_menus)
end
num_hist = length(history);
hist_menus = zeros(1,num_hist);
for i = 1:num_hist
hist_menus(i) = uimenu(hist_cm,'Label',history{i},...
'Callback',{cb,i});
end
end