max/ourMELONS
1
0
Fork 0
Code Issues 1 Pull requests Projects Releases Packages Wiki Activity Actions
ourMELONS/matlab/parallel/linkage_admix_parallel.m

1030 lines
33 KiB
Mathematica
Raw Normal View History

Added source Matlab code for reference
2019-12-16 16:47:21 +01:00
function linkage_admix_parallel(tietue, options)
global COUNTS; global PARTITION; global SUMCOUNTS;
clearGlobalVars;
PARTITION = tietue.PARTITION;
COUNTS = tietue.COUNTS;
SUMCOUNTS = tietue.SUMCOUNTS;
rowsFromInd = tietue.rowsFromInd;
data = double(tietue.data);
npops = tietue.npops;
noalle = tietue.noalle;
switch tietue.mixtureType
case 'linear_mix'
linkage_model = 'linear';
case 'codon_mix'
linkage_model = 'codon';
end
if isfield(tietue,'gene_lengths')
gene_lengths = tietue.gene_lengths;
else
[filename, pathname] = uigetfile('*.txt', 'Load file with lengths of the genes (same order as in data).');
gene_lengths = load([pathname filename]);
end
if length(unique(PARTITION(find(PARTITION>0))))==1
disp('Only one population in the input file');
disp('No admixture detected');
return
end
%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
%alaRaja = str2double(answers{1,1});
alaRaja = options.minSize;
[npops] = poistaLiianPienet(npops, rowsFromInd, alaRaja);
nloci = size(COUNTS,2);
ninds = size(data,1)/rowsFromInd;
iterationCount = options.iters;
nrefIndsInPop = options.refInds;
iterationCountRef = options.refIters;
[cliq_data, sep_data, cliq_counts, component_mat] = createCliqData(data, gene_lengths, noalle, ...
linkage_model, rowsFromInd);
% Repeat: simulate clique frequencies and estimate proportions. Save the
% average proportions in "proportionsIt".
proportionsIt = zeros(ninds,npops);
for iterationNum = 1:iterationCount
disp(['Iter: ' num2str(iterationNum)]);
%allfreqs = simulateAllFreqs(noalle);
[cliq_freqs, sep_freqs] = simulateCliqFreqs(cliq_counts, noalle, component_mat, gene_lengths, linkage_model);
for ind=1:ninds
%omaFreqs = computePersonalAllFreqs(ind, data, allfreqs, rowsFromInd);
[ownCliqFreqs, ownSepFreqs] = computePersonalCliqueFreqs(ind, ...
cliq_data, cliq_freqs, sep_data, sep_freqs, rowsFromInd, gene_lengths, linkage_model);
osuusTaulu = zeros(1,npops);
if PARTITION(ind)==0
% Outlier individual
elseif PARTITION(ind)~=0
if PARTITION(ind)>0
osuusTaulu(PARTITION(ind)) = 1;
else
% Individuals who are not assigned to any cluster.
arvot = zeros(1,npops);
for q=1:npops
osuusTaulu = zeros(1,npops);
osuusTaulu(q) = 1;
arvot(q) = computeIndLikelihood(ownCliqFreqs, ownSepFreqs, osuusTaulu);
end
[iso_arvo, isoimman_indeksi] = max(arvot);
osuusTaulu = zeros(1,npops);
osuusTaulu(isoimman_indeksi) = 1;
PARTITION(ind)=isoimman_indeksi;
end
logml = computeIndLikelihood(ownCliqFreqs, ownSepFreqs, osuusTaulu);
for osuus = [0.5 0.25 0.05 0.01]
[osuusTaulu, logml] = searchBest(osuus, osuusTaulu, ownCliqFreqs, ownSepFreqs, 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 t<EFBFBD>llaista.
%refData = simulateIndividuals(nrefIndsInPop,rowsFromInd,allfreqs);
exp_cliq_freqs = ...
computeExpectedFreqs(cliq_counts, noalle, component_mat, gene_lengths, linkage_model);
[ref_cliq_data, ref_sep_data] = ...
simulateLinkageIndividuals(nrefIndsInPop, rowsFromInd, exp_cliq_freqs, ...
gene_lengths, noalle, component_mat, linkage_model);
nrefInds = npops*nrefIndsInPop;
disp(['Analysing the reference individuals in ' num2str(iterationCountRef) ' iterations.']);
refProportions = zeros(nrefInds,npops);
for iter = 1:iterationCountRef
disp(['Iter: ' num2str(iter)]);
%allfreqs = simulateAllFreqs(noalle);
[cliq_freqs, sep_freqs] = simulateCliqFreqs(cliq_counts, noalle, component_mat, gene_lengths, linkage_model);
for ind = 1:nrefInds
%omaFreqs = computePersonalAllFreqs(ind, refData, allfreqs, rowsFromInd);
[ownCliqFreqs, ownSepFreqs] = computePersonalCliqueFreqs(ind, ...
ref_cliq_data, cliq_freqs, ref_sep_data, sep_freqs, rowsFromInd, gene_lengths, linkage_model);
osuusTaulu = zeros(1,npops);
pop = ceil(ind/nrefIndsInPop);
osuusTaulu(pop)=1;
%logml = computeIndLogml(omaFreqs, osuusTaulu);
logml = computeIndLikelihood(ownCliqFreqs, ownSepFreqs, osuusTaulu);
for osuus = [0.5 0.25 0.05 0.01]
%[osuusTaulu, logml] = searchBest(osuus, osuusTaulu, omaFreqs, logml);
[osuusTaulu, logml] = searchBest(osuus, osuusTaulu, ownCliqFreqs, ownSepFreqs, logml);
end
refProportions(ind,:) = refProportions(ind,:).*(iter-1) + osuusTaulu;
refProportions(ind,:) = refProportions(ind,:)./iter;
end
end
refTaulu = zeros(npops,100);
for ind = 1:nrefInds
pop = ceil(ind/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)) = refTaulu(pop, round(omanOsuus*100))+1;
end
% Rounding:
proportionsIt = proportionsIt.*100; proportionsIt = round(proportionsIt);
proportionsIt = proportionsIt./100;
for ind = 1:ninds
% if sum not equal to one, fix the largest part.
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
% "p-value" for admixture
uskottavuus = zeros(ninds,1);
for ind = 1:ninds
pop = PARTITION(ind);
if pop==0 % an outlier
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
refPienempia = sum(refTaulu(pop, 1:round(100*omanOsuus)));
uskottavuus(ind) = refPienempia / nrefIndsInPop;
end
end
tulostaAdmixtureTiedot(proportionsIt, uskottavuus, alaRaja, iterationCount);
%viewPartition(proportionsIt, popnames);
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
% copy 'baps4_output.baps' into the text file with the same name.
if exist('baps4_output.baps','file')
copyfile('baps4_output.baps',[options.outputMat '.txt'])
delete('baps4_output.baps')
end
tietue.proportionsIt = proportionsIt;
tietue.pvalue = uskottavuus; % Added by Jing
tietue.admixnpops = npops;
tietue.mixtureType = 'admix'; % added by jing on 09.09.2008
% save(options.outputMat, 'tietue');
save(options.outputMat, 'tietue','-v7.3'); % added by Lu Cheng, 08.06.2012
%----------------------------------------------------------------------------
function [npops] = poistaLiianPienet(npops, rowsFromInd, alaraja)
% Muokkaa tulokset muotoon, jossa outlier yksil<EFBFBD>t on
% poistettu. Tarkalleen ottaen poistaa ne populaatiot,
% joissa on v<EFBFBD>hemm<EFBFBD>n kuin 'alaraja':n verran yksil<EFBFBD>it?
global PARTITION;
global COUNTS;
global SUMCOUNTS;
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];
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;
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)
% Lis<EFBFBD><EFBFBD> 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 refData = simulateIndividuals(n,rowsFromInd,allfreqs)
% simuloidaan n yksil<EFBFBD><EFBFBD> jokaisesta populaatiosta. (
npops = size(allfreqs,3);
nloci = size(allfreqs,2);
ninds = n*npops;
refData = zeros(ninds*rowsFromInd,nloci);
counter = 1; % Pit<EFBFBD><EFBFBD> kirjaa mille riville seuraavaksi simuloidaan.
for ind = 1:ninds
pop = ceil(ind/n);
for loc = 1:nloci
for k=0:rowsFromInd-1
refData(counter+k,loc) = simuloiAlleeli(allfreqs,pop,loc);
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 loggis = computeIndLikelihood(ownCliqFreqs, ownSepFreqs, proportions)
% Calculates the likelihood when the origins are defined by "proportions".
aux = proportions * ownCliqFreqs;
aux = log(aux);
loggis = sum(aux);
clear aux;
aux2 = proportions * ownSepFreqs;
aux2 = log(aux2);
loggis = loggis - sum(aux2);
%--------------------------------------------------------------------------
function osuusTaulu = suoritaMuutos(osuusTaulu, osuus, indeksi)
% P<EFBFBD>ivitt<EFBFBD><EFBFBD> osuusTaulun muutoksen j<EFBFBD>lkeen.
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] = searchBest(osuus, osuusTaulu, ownCliqFreqs, ownSepFreqs, logml)
ready = 0;
while ready ~= 1
muutokset = calcChanges(osuus, osuusTaulu, ownCliqFreqs, ownSepFreqs, 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 = calcChanges(osuus, osuusTaulu, ownCliqFreqs, ownSepFreqs, logml)
% Palauttaa npops*npops taulun, jonka alkio (i,j) kertoo, mik?on
% muutos logml:ss? mik<EFBFBD>li populaatiosta i siirret<EFBFBD><EFBFBD>n osuuden verran
% todenn<EFBFBD>k<EFBFBD>isyysmassaa populaatioon j. Mik<EFBFBD>li populaatiossa i ei ole
% mit<EFBFBD><EFBFBD>n siirrett<EFBFBD>v<EFBFBD><EFBFBD>, 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 = computeIndLikelihood(ownCliqFreqs, ownSepFreqs, osuusTaulu);
muutokset(i1,i2) = loggis-logml;
osuusTaulu(i2) = osuusTaulu(i2)-osuus;
end
osuusTaulu(i1) = osuusTaulu(i1)+osuus;
end
%---------------------------------------------------------------
function dispLine
disp('---------------------------------------------------');
%--------------------------------------------------------------------------
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
%------------------------------------------------------
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;
%-------------------------------------------------------
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)
% K<EFBFBD>ytt<EFBFBD>esim 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 waitALittle
A = rand(500);
gammaln(A);
%--------------------------------------------------
function [cliq_data, sep_data, cliq_counts, component_mat] = ...
createCliqData(data, gene_lengths, noalle, linkage_model, ...
rowsFromInd)
% cliq_data: cell array, each cell corresponds to one gene. Element (i,j)
% in cell k is the code of the allele combination in the j:th clique in
% gene k, for individual i.
% sep_data: like cliq_data. i:th separator separates cliques i and i+1.
% cliq_counts: cell array, each cell corresponds to one gene. Each cell is
% a 3-dimensional array, where the element (i,j,k) is the observed count of
% allele combination i, in clique j, in population k.
% The coding of the allele combinations: If a clique of 3 sites has
% noalle values 3,2,4, then the allele combinations are given numbers in
% lexicographic order: 111, 112, 113, 114, 121, 122, ..., 324.
%------------------------------------------------------------------------
% cliq_data on cell-array, jossa kukin solu vastaa yht?geeni? Alkio (i,j)
% solussa k merkitsee sen alleelikombinaation koodia, joka yksil<EFBFBD>ll?i
% havaitaan geenin k klikiss?numero j.
% cliq_counts on cell-array, jossa my<EFBFBD>s kukin solu vastaa yht?geeni?
% Kukin solu on kolmiulotteinen taulukko, jonka alkio (i,j,k) on
% populaatiossa k, ko geenin j:nness?klikiss?havaitun alleelikombinaation
% i lukum<EFBFBD><EFBFBD>r?
% Alleelikombinaatioiden koodaus: Jos kolmen position klikiss?on (koko
% datassa) noalle:t 3,2,4, (eli ekassa positiossa alleelit 1-3, tokassa 1-2
% ja kolmannessa alleelit 1-4), niin alleelikombinaatiot numeroidaan
% leksikograafisessa j<EFBFBD>rjestyksess? 111, 112, 113, 114, 121, 122, ...,
% 324.
%-----------------------------------------------------------------------
global PARTITION;
if sum(gene_lengths) ~= size(data,2)
disp('Error 155');
end
if ~isa(data,'double')
data = double(data); % Required in matlab 6
end
ninds = size(data,1);
n_genes = length(gene_lengths);
if strcmp(linkage_model,'codon')
n_cliques = gene_lengths-2; % Number of cliques in each gene
else
n_cliques = gene_lengths-1; % Use linear model
end
max_noalle = zeros(n_genes,1); % Maximum "clique noalle" in each gene.
component_mat = zeros(n_genes, max(gene_lengths));
cum_length = cumsum(gene_lengths);
component_mat(1,1:gene_lengths(1))=1:gene_lengths(1);
for i = 2:n_genes
component_mat(i,1:gene_lengths(i)) = cum_length(i-1)+1:cum_length(i);
end
for i = 1:n_genes
% What is the largest number of different values that are observed for
% some clique in this gene:
number = 0;
if n_cliques(i)<1
% the gene is shorter than a normal clique.
if gene_lengths(i)==2
% linkage_model must be 'codon' to end up here..
positions = component_mat(i, [1 2]);
number = prod(noalle(positions));
else
% gene_lengths(i) == 1
positions = component_mat(i,1);
number = noalle(positions);
end
else
for j = 1:n_cliques(i)
if strcmp(linkage_model,'codon'), positions = component_mat(i , j:j+2);
else positions = component_mat(i, j:j+1);
end
cand = prod(noalle(positions));
if cand>number
number=cand;
end
end
end
max_noalle(i) = number;
end
cliq_data = cell(n_genes,1); % An array for each gene.
% (i,j) is the combination which individual i has in clique j (in haploid case..).
for i = 1:n_genes
if n_cliques(i)<1
cliq_data{i} = zeros(ninds, 1);
if gene_lengths(i)==2
% linkage_model must be 'codon' to end up here..
positions = component_mat(i, [1 2]);
rows = data(:,positions);
observations = (rows(:,1)-1) * noalle(positions(2)) + rows(:,2);
else
positions = component_mat(i,1);
rows = data(:,positions);
observations = rows;
end
cliq_data{i}(:,1) = observations;
else
cliq_data{i} = zeros(ninds, n_cliques(i));
for j = 1:n_cliques(i)
if strcmp(linkage_model,'codon')
positions = component_mat(i,j:j+2);
rows = data(:, positions);
observations = (rows(:,1)-1) * prod(noalle(positions(2:3))) + ...
(rows(:,2)-1) * noalle(positions(3)) + rows(:,3);
else
positions = component_mat(i,j:j+1);
rows = data(:,positions);
observations = (rows(:,1)-1) * noalle(positions(2)) + rows(:,2);
end
cliq_data{i}(:,j) = observations;
end
end
end
cliq_counts = cell(n_genes,1);
% (i,j,k) is the count of combination i, in clique j, in population k.
npops = length(unique(PARTITION));
for i = 1:n_genes
cliq_counts{i} = zeros(max_noalle(i), max(1,n_cliques(i)), npops);
for j = 1:npops
partition = repmat(PARTITION', [rowsFromInd 1]);
partition = partition(:); % Partition for rows in the data (instead of individuals).
inds_now = find(partition==j);
ninds_now = length(inds_now);
data_now = cliq_data{i}(inds_now,:);
for k = 1:max(n_cliques(i),1)
apu = zeros(ninds_now, max_noalle(i));
apu(sub2ind([ninds_now max_noalle(i)],...
(1:ninds_now)', data_now(:,k)))=1;
cliq_counts{i}(:, k, j) = (sum(apu,1))';
end
end
end
sep_data = cell(n_genes,1);
n_separators = n_cliques-1;
for i = 1:n_genes
sep_data{i} = zeros(ninds, n_separators(i));
for j = 1:n_separators(i)
if strcmp(linkage_model, 'codon')
positions = component_mat(i,j+1:j+2);
rows = data(:, positions);
observations = (rows(:,1)-1) * noalle(positions(2)) + rows(:,2);
else
positions = component_mat(i,j+1);
rows = data(:,positions);
observations = rows;
end
sep_data{i}(:,j) = observations;
end
end
%------------------------------------------------------
function [cliq_freqs, sep_freqs] = simulateCliqFreqs(cliq_counts, noalle, component_mat, ...
gene_lengths, linkage_model)
% cliq_freqs: cell-array. Element (i,j,k) in cell m is the frequence of
% combination i, in clique j of the m:th gene, in population k.
% sep_freqs: like cliq_freqs, but for the separators.
%------------------------------------------------------------------------
% cliq_freqs: cell-array, jossa on vastaavat dimensiot kuin cliq_counts:issa.
% solun m alkio (i,j,k) on geenin m, klikin j, kombinaation i, frekvenssi
% populaatiossa k.
% sep_freqs: cell-array, kuten cl_freqs, mutta separaattoreille.
%-------------------------------------------------------------------------
global PARTITION;
n_genes = length(cliq_counts);
if strcmp(linkage_model,'codon')
n_cliques = gene_lengths-2; % Number of cliques in each gene
else
n_cliques = gene_lengths-1; % Use linear model
end
npops = length(unique(PARTITION));
cliq_freqs = cell(n_genes,1);
sep_freqs = cell(n_genes,1);
for i=1:n_genes
cliq_freqs{i} = zeros(size(cliq_counts{i}));
positions = component_mat(i,1:gene_lengths(i));
if n_cliques(i)<1
% the gene is shorter than a normal clique.
if gene_lengths(i)==2
% linkage_model must be 'codon' to end up here..
cliq_noalle = noalle(positions(1)) .* noalle(positions(2));
sep_noalle = [];
else
% gene_lengths(i) == 1
cliq_noalle = noalle(positions(1));
sep_noalle = [];
end
sep_freqs{i} = [];
else
if strcmp(linkage_model, 'codon')
cliq_noalle = noalle(positions(1:end-2)) .* noalle(positions(2:end-1)) .* ...
noalle(positions(3:end));
sep_noalle = noalle(positions(2:end-2)) .* noalle(positions(3:end-1));
else
cliq_noalle = noalle(positions(1:end-1)) .* noalle(positions(2:end));
sep_noalle = noalle(positions(2:end-1));
end
sep_freqs{i} = zeros(max(sep_noalle), n_cliques(i)-1, npops);
end
% First clique:
prior = (1 / cliq_noalle(1)) * ones(cliq_noalle(1),1);
counts_now = repmat(prior, [1 1 npops]) + cliq_counts{i}(1:cliq_noalle(1),1,:);
for k = 1:npops
simul = randdir(counts_now(:,1,k), cliq_noalle(1));
cliq_freqs{i}(1:cliq_noalle(1),1,k) = simul;
end
for j=2:n_cliques(i)
% Obtain freqs for j-1:th separator by marginalization from j-1:th
% clique, and draw values for the frequencies of the j:th clique:
aux = cliq_freqs{i}(1:cliq_noalle(j-1), j-1, :); % Freqs of the previous clique
aux = reshape(aux, [sep_noalle(j-1), noalle(positions(j-1)), npops]);
% Freqs for separator by marginalization from the previous clique:
sep_freqs{i}(1:sep_noalle(j-1),j-1,:) = sum(aux,2);
prior = (1 / cliq_noalle(j)) * ones(cliq_noalle(j),1);
counts_now = repmat(prior, [1 1 npops]) + cliq_counts{i}(1:cliq_noalle(j),j,:);
for k = 1:npops
% Simulate conditional frequencies:
for m = 1:sep_noalle(j-1)
if strcmp(linkage_model, 'codon')
values = (m-1)*noalle(positions(j+2))+1:m*noalle(positions(j+2));
else
values = (m-1)*noalle(positions(j+1))+1:m*noalle(positions(j+1));
end
simul = randdir(counts_now(values,1,k), length(values));
cliq_freqs{i}(values,j,k) = simul * sep_freqs{i}(m,j-1,k); % MIETI TARKKAAN!
end
end
end
end
%--------------------------------------------------------------------
function [ownCliqFreqs, ownSepFreqs] = computePersonalCliqueFreqs(...
ind, cl_data, cl_freqs, sep_data, sep_freqs, rowsFromInd, ...
gene_lengths, linkage_model)
% ownCliqFreqs is (npops * (n_cliques*rowsFromInd)) table, where each column
% contains the frequencies of the corresponding clique_combination, in
% different populations.
% ownSepFreqs is (npops * (n_seps*rowsFromInd)) table, like ownCliqFreqs.
n_genes = length(gene_lengths);
if strcmp(linkage_model,'codon')
n_cliques = gene_lengths-2;
n_cliques = max([n_cliques ones(n_genes,1)], [], 2); % for genes shorter than clique
else
n_cliques = gene_lengths-1; % Use linear model.
n_cliques = max([n_cliques ones(n_genes,1)], [], 2);
end
total_n_cliques = sum(n_cliques);
npops = size(cl_freqs{1},3);
ownCliqFreqsXX = zeros(1, total_n_cliques*rowsFromInd, npops);
pointer = 1;
for i = 1:n_genes
ind_data = cl_data{i}((ind-1)*rowsFromInd+1:ind*rowsFromInd , :);
for j = 1:n_cliques(i) % MUUTA!
for k = 1:rowsFromInd
code = ind_data(k,j);
ownCliqFreqsXX(1,pointer,:) = cl_freqs{i}(code,j,:);
pointer = pointer+1;
end
end
end
ownCliqFreqs = (squeeze(ownCliqFreqsXX))';
n_separators = n_cliques-1;
total_n_separators = sum(n_separators);
ownSepFreqsXX = zeros(1, total_n_separators*rowsFromInd, npops);
pointer = 1;
for i = 1:n_genes
ind_data = sep_data{i}((ind-1)*rowsFromInd+1:ind*rowsFromInd , :);
for j = 1:n_separators(i)
for k = 1:rowsFromInd
code = ind_data(k,j);
ownSepFreqsXX(1,pointer,:) = sep_freqs{i}(code,j,:);
pointer = pointer+1;
end
end
end
if (total_n_separators*rowsFromInd)==1
ownSepFreqs = (squeeze(ownSepFreqsXX));
else
ownSepFreqs = (squeeze(ownSepFreqsXX))';
end
%-------------------------------------------------------------------------
function exp_cliq_freqs = computeExpectedFreqs(cliq_counts, ...
noalle, component_mat, gene_lengths, linkage_model)
% Returns the expected values for the clique and separator frequencies in
% different populations.
% exp_cliq_freqs: cell-array. Element (i,j,k) in cell m is the expected
% frequence of combination i, in clique j of the m:th gene, in
% population k.
n_genes = length(gene_lengths);
if strcmp(linkage_model, 'codon')
n_cliques = gene_lengths-2;
else
n_cliques = gene_lengths-1; % Linear model
end
npops = size(cliq_counts{1},3);
exp_cliq_freqs = cell(n_genes,1);
for i = 1:n_genes
exp_cliq_freqs{i} = zeros(size(cliq_counts{i}));
positions = component_mat(i,1:gene_lengths(i));
if n_cliques(i)<1
% the gene is shorter than a normal clique.
if gene_lengths(i)==2
% linkage_model must be 'codon' to end up here..
cliq_noalle = noalle(positions(1)) .* noalle(positions(2));
else
% gene_lengths(i) == 1
cliq_noalle = noalle(positions(1));
end
else
if strcmp(linkage_model, 'codon')
cliq_noalle = noalle(positions(1:end-2)) .* noalle(positions(2:end-1)) .* ...
noalle(positions(3:end));
else
cliq_noalle = noalle(positions(1:end-1)) .* noalle(positions(2:end));
end
end
for j = 1:max(1, n_cliques(i))
prior = (1 / cliq_noalle(j)) * ones(cliq_noalle(j),1);
counts_now = repmat(prior, [1 1 npops]) + cliq_counts{i}(1:cliq_noalle(j),j,:);
exp_cliq_freqs{i}(1:cliq_noalle(j),j,:) = ...
counts_now ./ repmat(sum(counts_now,1), [cliq_noalle(j) 1 1]);
end
end
%----------------------------------------------------------
function [ref_cliq_data, ref_sep_data] = ...
simulateLinkageIndividuals(n, rowsFromInd, exp_cliq_freqs, gene_lengths, ...
noalle, component_mat, linkage_model)
% Simulates n individuals from each population using expected frequencies
% for cliques and separators.
% ref_cliq_data: cell array, each cell corresponds to one gene. Elements
% ((i-1)*rowsFromInd+1:i*rowsFromInd, j) in cell k are the codes of the allele
% combinations in the j:th clique in gene k, for individual i.
% ref_sep_data: like ref_cliq_data. i:th separator separates cliques i and i+1.
n_genes = length(gene_lengths);
if strcmp(linkage_model,'codon')
n_cliques = gene_lengths-2;
else
n_cliques = gene_lengths-1; % Linear model
end
npops = size(exp_cliq_freqs{1},3);
ninds = n*npops;
ref_cliq_data = cell(n_genes,1);
ref_sep_data = cell(n_genes,1);
for i = 1:n_genes
ref_cliq_data{i} = zeros(ninds*rowsFromInd, max(n_cliques(i),1)); % Added: rowsFromInd
positions = component_mat(i,1:gene_lengths(i));
if strcmp(linkage_model,'codon')
sep_noalle = noalle(positions(2:end-2)) .* noalle(positions(3:end-1));
else
sep_noalle = noalle(positions(2:end-1));
end
ref_sep_data{i} = zeros(ninds*rowsFromInd, n_cliques(i)-1); % Added: rowsFromInd
for ind = 1:ninds
pop = ceil(ind/n);
% First clique:
freqs = exp_cliq_freqs{i}(:,1,pop);
freqs = repmat(freqs, [1 rowsFromInd]);
codes = simulateCodes(freqs);
ref_cliq_data{i}((ind-1)*rowsFromInd+1:ind*rowsFromInd, 1) = codes;
for j = 2:n_cliques(i)
previous_cliq = ref_cliq_data{i}((ind-1)*rowsFromInd+1:ind*rowsFromInd, j-1);
% Value for j-1:th separator:
sep_codes = rem(previous_cliq, sep_noalle(j-1));
sep_codes(find(sep_codes==0)) = sep_noalle(j-1);
ref_sep_data{i}((ind-1)*rowsFromInd+1:ind*rowsFromInd, j-1) = sep_codes;
% Value for j:th clique:
if strcmp(linkage_model,'codon')
freqs = zeros(noalle(positions(j+2)),rowsFromInd);
values = zeros(noalle(positions(j+2)),rowsFromInd);
for k = 1:rowsFromInd
values(:,k) = ((sep_codes(k)-1)*noalle(positions(j+2))+1 : sep_codes(k)*noalle(positions(j+2)))';
freqs(:,k) = exp_cliq_freqs{i}(values(:,k), j, pop);
end
freqs = freqs ./ repmat(sum(freqs,1), [noalle(positions(j+2)) 1]);
else
freqs = zeros(noalle(positions(j+1)),rowsFromInd);
values = zeros(noalle(positions(j+1)),rowsFromInd);
for k = 1:rowsFromInd
values(:,k) = ((sep_codes(k)-1)*noalle(positions(j+1))+1 : sep_codes(k)*noalle(positions(j+1)))';
freqs(:,k) = exp_cliq_freqs{i}(values(:,k), j, pop);
end
freqs = freqs ./ repmat(sum(freqs,1), [noalle(positions(j+1)) 1]);
end
codes = simulateCodes(freqs);
codes = values(sub2ind(size(values),codes,(1:rowsFromInd)'));
ref_cliq_data{i}((ind-1)*rowsFromInd+1:ind*rowsFromInd, j) = codes;
end
end
end
function codes = simulateCodes(freqs)
% Freqs is a table where each column is a distribution. The
% number of columns in freqs must be equal to rowsFromInd.
% A value is drawn from each distribution in different columns. The values
% are saved in codes, which is (rowsFromInd*1) table.
[nrows, rowsFromInd] = size(freqs);
codes = nrows+1-sum(cumsum(freqs)>repmat(rand(1,rowsFromInd),[nrows 1]),1);
codes = codes';
Reference in a new issue Copy permalink