config file added. comik_wrapper now accepts a config file directly. …

…Minor edit in getExpansionPoints

comik_wrapper.m

@@ -1,11 +1,13 @@
-function [] = comik_wrapper(givenPosFastaFilename, givenNegFastaFilename, nPosSequences, nNegSequences, testIndices, outputFolder, oligoLen, maxDist, segmentSizeInBps, nClusterVals, sigmaVals, Cs, mklNorm, nFolds, nOuterFolds, whetherToPlotHeatmap, whetherToVisualizeWVector, debugLevel, debugMsgLocation, computationVersion)
+function [] = comik_wrapper(configFile)
 
 % COMIK_WRAPPER
 % Usage:
-%     comik_wrapper(givenFastaFilename, nPosSequences, nNegSequences, ...
-%     outputFolder, oligoLen, maxDist, segmentSizeInBps, nClusterVals, ...
-%     sigmaVals, Cs, mklNorm, nFolds, nOuterFolds, whetherToPlotHeatmap, ...
-%     computationVersion, whetherToVisualizeWVector)
+%     comik_wrapper(configFile)
+% 
+% INPUT PARAMS
+% Param 'configFile'
+% Config filename with the param values set. More information on the params 
+% given below
 %
 % Only the following params have default values:
 %
@@ -20,9 +22,8 @@
 % debugMsgLocation		1
 % computationVersion		'Looping'
 % 
-% Required params:
-% ----------------
-% INPUT PARAMS
+% Rest, params that are required to be specified:
+% -----------------------------------------------
 % Param 'givenFastFilename' (string)
 % The input FASTA file should contain all sequences together, positives 
 % followed by negatives.
@@ -63,146 +64,35 @@
 % 
 % Author: snikumbh@mpi-inf.mpg.de
 
-totalArguments = 20;
-if nargin < totalArguments
-    computationVersion = 'Looping';
-end
-if nargin < totalArguments - 1
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-end
-if nargin < totalArguments - 2
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-end
-if nargin < totalArguments - 3
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-end
-if nargin < totalArguments - 4
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-end
-if nargin < totalArguments - 5
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-end
-if nargin < totalArguments - 6
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-end
-if nargin < totalArguments - 7
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-end
-if nargin < totalArguments - 8
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-    Cs = 10.^[-3:1:3];
-end
-if nargin < totalArguments - 9
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-    Cs = 10.^[-3:1:3];
-    sigmaVals = 10.^[-1:1:2];
-end
-if nargin < totalArguments - 10
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-    Cs = 10.^[-3:1:3];
-    sigmaVals = 10.^[-1:1:2];
-    nClusterVals = [2 5];
-end
-if nargin < totalArguments - 11
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-    Cs = 10.^[-3:1:3];
-    sigmaVals = 10.^[-1:1:2];
-    nClusterVals = [2 5];
-    segmentSizeInBps = 100;
-end
-if nargin < totalArguments - 12
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-    Cs = 10.^[-3:1:3];
-    sigmaVals = 10.^[-1:1:2];
-    nClusterVals = [2 5];
-    segmentSizeInBps = 100;
-    maxDist = 50;
-end
-if nargin < totalArguments - 13
-    computationVersion = 'Looping';
-    debugMsgLocation = 1;
-    debugLevel = 2;
-    whetherToVisualizeWVector = 'Yes';
-    whetherToPlotHeatmap = 'No';
-    nOuterFolds = 5;
-    nFolds = 10;
-    mklNorm = 2.0;
-    Cs = 10.^[-3:1:3];
-    sigmaVals = 10.^[-1:1:2];
-    nClusterVals = [2 5];
-    segmentSizeInBps = 100;
-    maxDist = 50;
-    oligoLen = [2];
-end
+if nargin == 1
+    % check, this should be the config filename
+    % default name: config-comik.txt
+    if ischar(configFile)
+	if exist(configFile, 'file') == 2
+	    fid = fopen(configFile, 'r');
+	    l = fgetl(fid);
+	    firstLineConfigFile = '## CoMIK CONFIG FILE';
+	    if findstr(l, firstLineConfigFile) > 0
+	        % configParams is a struct
+	        configParams = readConfigFile(configFile);
+	    end
+            fclose(fid);
+	else
+	    eMsg = 'Given config file does not exist!';
+	    error(eMsg);
+	end
+    else
+	error('Error.\nInput must be a char, not a %s.', class(configFile));
+    end
+else
+    error('Specify only the config filename!');
+end 
 
 
-if exist(outputFolder, 'dir') == 7
+if exist(configParams.outputFolder, 'dir') == 7
     %do-nothing
 else
-    status = mkdir(outputFolder);
+    status = mkdir(configParams.outputFolder);
 end
 
 % -- Perform different outer folds with parfor, asynchronously
@@ -211,13 +101,13 @@
 
 % Set seed for the random number generator
 rng(21, 'twister');
-posIndices = randperm(nPosSequences);
-negIndices = randperm(nNegSequences);
+posIndices = randperm(configParams.nPosSequences);
+negIndices = randperm(configParams.nNegSequences);
 
-posTestIndices = intersect(testIndices, posIndices);
-negTestIndices = setdiff(testIndices, posIndices) - nPosSequences;
+posTestIndices = intersect(configParams.testIndices, posIndices);
+negTestIndices = setdiff(configParams.testIndices, posIndices) - configParams.nPosSequences;
 
-thisTestIndices{1} = testIndices;
+thisTestIndices{1} = configParams.testIndices;
 for i=2:nOuterFolds
     thisPosTestIndices = circshift(posIndices, [0, (i-1)*length(posTestIndices)]);
     thisNegTestIndices = circshift(negIndices, [0, (i-1)*length(negTestIndices)]) + length(posIndices);
@@ -226,25 +116,26 @@
 %% Conditionally create a new pool; delete any existing pool of workers if NumWorkers is less than required, else use the same poolobj
 poolobj = gcp('nocreate');
 if isempty(poolobj)
-    parpool('local', nOuterFolds);
+    parpool('local', configParams.nOuterFolds);
 else
-    if poolobj.NumWorkers >= nOuterFolds
-	logMessages(1, sprintf('Reusing existing parallel pool: %d out of %d workers\n', nOuterFolds, poolobj.NumWorkers), debugLevel);
+    if poolobj.NumWorkers >= configParams.nOuterFolds
+	logMessages(1, sprintf('Reusing existing parallel pool: %d out of %d workers\n', configParams.nOuterFolds, poolobj.NumWorkers), configParams.debugLevel);
     else
-        logMessages(1, sprintf('Not enough workers in the existing parallel pool.\nDeleting existing pool with %d workers\n', poolobj.NumWorkers), debugLevel);
+        logMessages(1, sprintf('Not enough workers in the existing parallel pool.\nDeleting existing pool with %d workers\n', poolobj.NumWorkers), configParams.debugLevel);
         delete(poolobj);
-        parpool('local', nOuterFolds);
+        parpool('local', configParams.nOuterFolds);
     end
 end
-parfor (i=1:nOuterFolds, nOuterFolds)
-    thisFoldDirectory = strcat(outputFolder, '/outer_fold_', num2str(i));
-    returnStatusVal = perform_one_outer_fold_for_comik(i, givenPosFastaFilename, givenNegFastaFilename, nPosSequences, nNegSequences, oligoLen, ...
-							maxDist, segmentSizeInBps, nClusterVals, sigmaVals, Cs, mklNorm, nFolds, ...
-							thisTestIndices{i}, debugLevel, thisFoldDirectory, whetherToPlotHeatmap, ...
-							computationVersion, whetherToVisualizeWVector, debugMsgLocation);
-    logMessages(1, sprintf('\nOuter fold %d: Status %s\n', i, returnStatusVal), debugLevel);
+parfor (i=1:configParams.nOuterFolds, configParams.nOuterFolds)
+    thisFoldDirectory = strcat(configParams.outputFolder, '/outer_fold_', num2str(i));
+    returnStatusVal = perform_one_outer_fold_for_comik(i, configParams.givenPosFastaFilename, configParams.givenNegFastaFilename, configParams.nPosSequences, ...
+							configParams.nNegSequences, configParams.oligoLen, configParams.maxDist, configParams.segmentSizeInBps, ...
+							configParams.nClusterVals, configParams.sigmaVals, configParams.Cs, configParams.mklNorm, configParams.nFolds, ...
+							thisTestIndices{i}, configParams.debugLevel, thisFoldDirectory, configParams.whetherToPlotHeatmap, ...
+							configParams.computationVersion, configParams.whetherToVisualizeWVector, configParams.debugMsgLocation);
+    logMessages(1, sprintf('\nOuter fold %d: Status %s\n', i, returnStatusVal), configParams.debugLevel);
 end
-logMessages(1, sprintf('All outer folds completed!\n'), debugLevel);
+logMessages(1, sprintf('All outer folds completed!\n'), configParams.debugLevel);
 
 
 end % comik_wrapper function ends

config-comik.txt

@@ -0,0 +1,29 @@
+## CoMIK CONFIG FILE
+
+## Required Input
+POSITIVE_FASTA_FILE=./sample_data/simulated_dataset1/pos.fasta
+NEGATIVE_FASTA_FILE=./sample_data/simulated_dataset1/neg.fasta
+NUMBER_OF_POSITIVES=600
+NUMBER_OF_NEGATIVES=600
+TEST_INDICES=[501:600 1101:1200]
+OUTPUT_FOLDER=comik_run_simulated_dataset1
+
+##  ODH requirements
+OLIGO_LEN=[2 3]
+MAX_DIST=10
+
+##  For CoMIK
+SEGMENT_SIZE_IN_BPS=10
+NUMBER_OF_CLUSTERS=[2 5 7]
+SIGMA_VALUES=10.^[1:1:2]
+COST_VALUES=10.^[-3:1:3]
+## Above are params which are required
+## Further, params with otherwise default values
+MKL_NORM=2.0
+NUMBER_OF_INNER_FOLDS=10
+NUMBER_OF_OUTER_FOLDS=5
+WHETHER_TO_PLOT_HEATMAP=No
+WHETHER_TO_VISUALIZE_WEIGHT_VECTOR=Yes
+DEBUG_LEVEL=0
+DEBUG_MSG_LOCATION=runLog.txt
+COMPUTATION_VERSION=Looping

getExpansionPoints.m

@@ -35,7 +35,7 @@
 rng('default');
 logMessages(debugMsgLocation, sprintf('with K-means: \n'), debugLevel);
 if debugLevel == 2
-    [idx, expansionPoints, sumD, matrixOfDistancesFromCentres] = kmeans(X, nClusters, 'Replicates', rep, 'Display','on');%, 'MaxIter',1000);%Display: final
+    [idx, expansionPoints, sumD, matrixOfDistancesFromCentres] = kmeans(X, nClusters, 'Replicates', rep, 'Display','Iter');%, 'MaxIter',1000);%Display: final
 elseif debugLevel == 0
     [idx, expansionPoints, sumD, matrixOfDistancesFromCentres] = kmeans(X, nClusters, 'Replicates', rep, 'Display','off');%, 'MaxIter',1000);
 end