/**
* Called from bicat.gui.window.GenePairAnalysis
*
* @throws Exception
*/
public void genePairAnalysis(
int data, int list, int idx, int minCoocScore, int minCommonScore, boolean byCooc)
throws Exception {
Dataset BcR = (Dataset) engine.getDatasetList().get(data);
LinkedList biclusterList = (LinkedList) BcR.getBCsList(list).get(idx);
HashMap gpa = new HashMap();
if (engine.isDebug()) {
System.out.println("Starting gene pair analysis ...");
System.out.println("Hashmap for biclusterList: " + biclusterList.toString());
}
if (byCooc) gpa = gpaByCoocurrence(biclusterList, minCoocScore, BcR);
else gpa = gpaByCommonChips(biclusterList, minCommonScore);
if (engine.isDebug()) System.out.println("Hashmap size for gpa: " + gpa.size());
// do management ...
BcR.updateAnalysisBiclustersLists(gpa, list, idx);
if (owner != null) {
owner.updateAnalysisMenu();
owner.buildTree();
owner.getTree().setSelectionPath(owner.getPreprocessedPath());
owner.getTree().setSelectionPath(owner.getPreprocessedPath());
int row = owner.getTree().getRowCount() - 1;
while (row >= 0) {
owner.getTree().collapseRow(row);
row--;
}
}
}
/** Called from bicat.gui.window.Filter */
public void filter(
int data, int list, int idx, int minG, int maxG, int minC, int maxC, int nrBCs, int overlap) {
if (engine.isDebug()) System.out.println("Starting filter ...");
Dataset BcR = (Dataset) engine.getDatasetList().get(data);
LinkedList biclusterList = (LinkedList) BcR.getBCsList(list).get(idx);
int limit_G = BcR.getGeneCount();
int limit_C = BcR.getWorkingChipCount();
biclusterList = filterBySize(biclusterList, minG, maxG, minC, maxC, limit_G, limit_C);
biclusterList = filterByOverlap(biclusterList, nrBCs, overlap);
// do management ...
BcR.updateFilterBiclustersLists(biclusterList, list, idx);
if (owner != null) {
owner.updateFilterMenu();
owner.buildTree();
owner.getTree().setSelectionPath(owner.getPreprocessedPath());
int row = owner.getTree().getRowCount() - 1;
while (row >= 0) {
owner.getTree().collapseRow(row);
row--;
}
}
}
/** Called from bicat.gui.window.Search */
public void search(int data, int list, int idx, String genes, String chips, boolean andSearch) {
if (engine.isDebug()) System.out.println("Starting search ...");
Dataset BcR = (Dataset) engine.getDatasetList().get(data);
LinkedList biclusterList = (LinkedList) BcR.getBCsList(list).get(idx);
biclusterList = search(biclusterList, genes, chips, andSearch);
// do management ...
BcR.updateSearchBiclustersLists(biclusterList, list, idx);
if (owner != null) {
owner.updateSearchMenu();
owner.buildTree();
owner.getTree().setSelectionPath(owner.getPreprocessedPath());
owner.getTree().setSelectionPath(owner.getPreprocessedPath());
int row = owner.getTree().getRowCount() - 1;
while (row >= 0) {
owner.getTree().collapseRow(row);
row--;
}
}
}
/**
* First dim (data.length) is the number of items to cluster. Dementspreched is the second dim
* (data[0].length) number of attributes.
*
* <p>Add the labeling management. + Outputting (joins)
*
* <p>OUTPUT format: C.nr_joins.dist\n L x1 L x2
*
* <p>C.nr_joins.dist\n L x3 C N1 ...
*
* @param distMetr
*/
public void runAgglomerative(float[][] data, int distMetr) {
StringBuffer sb = new StringBuffer(); // at the end, write it OUT
distance_metric = distMetr;
int nr_genes = data.length;
int nr_joins = 0; // numbering of clusters begins @ 1
int[] back_translation = new int[nr_genes]; // current L idx -> original
Vector clusters = new Vector(); // contains extended view of the
// clusters, as they are created
if (DEBUG) System.out.println("nr_items = " + nr_genes);
// INIT: compute the distance matrix, and create the singletons:
double[][] distance_matrix = new double[nr_genes][nr_genes];
if (MethodConstants.debug) System.out.println("The distance metric is: " + distance_metric);
for (int i = 0; i < nr_genes; i++) {
distance_matrix[i][i] = 0.0;
for (int j = i + 1; j < nr_genes; j++) {
switch (distance_metric) {
case MethodConstants.EUCLIDEAN_DISTANCE:
distance_matrix[i][j] = Util.computeEuclideanDistance(data[i], data[j]);
break;
case MethodConstants.PEARSON_CORRELATION:
distance_matrix[i][j] = Util.computePearsonCorrelationDistance(data[i], data[j]);
break;
case MethodConstants.MANHATTAN_DISTANCE:
distance_matrix[i][j] = Util.computeManhattanDistance(data[i], data[j]);
break;
case MethodConstants.MINKOWSKI_DISTANCE:
distance_matrix[i][j] = Util.computeMinkowskiDistance(data[i], data[j]);
break;
case MethodConstants.COSINE_DISTANCE:
distance_matrix[i][j] = Util.computeCosineDistance(data[i], data[j]);
break;
default:
break; /* not foreseen */
}
distance_matrix[j][i] = distance_matrix[i][j];
if (MethodConstants.debug)
System.out.println(
"i: " + i + " j: " + j + " distance Matrix[i][j]: " + distance_matrix[i][j]);
}
}
Vector cluster_idxs = new Vector();
for (int i = 0; i < nr_genes; i++) {
int[] cl = new int[1];
cl[0] = i;
cluster_idxs.add(cl);
}
Vector cluster_dists = new Vector();
for (int i = 0; i < nr_genes; i++) {
Vector distance_vector = new Vector();
for (int j = 0; j < nr_genes; j++) distance_vector.add(new Double(distance_matrix[i][j]));
cluster_dists.add(distance_vector);
}
for (int i = 0; i < nr_genes; i++) {
back_translation[i] = i;
}
while (cluster_idxs.size() > N) {
// 2. merge (into) clusters
int[] idxs = findMinDistancePair(cluster_dists); // 2 (i,j)
// if(DEBUG) System.out.println("cl_dists size = " +
// cluster_dists.size());
int[] cl_1 = (int[]) cluster_idxs.get(idxs[0]);
int[] cl_2 = (int[]) cluster_idxs.get(idxs[1]);
if (DEBUG) {
if (idxs[0] >= idxs[1]) {
System.out.println("ATTENTION: idxs[0] >= idxs[1] ");
}
}
// --- Management Ops:
nr_joins++;
sb.append("C " + nr_joins + " " + curr_distance + " ");
if (((int[]) cluster_idxs.get(idxs[0])).length == 1) // sb.append("L
sb.append("L " + back_translation[idxs[0]] + " ");
else {
int[] cl_0 = ((int[]) cluster_idxs.get(idxs[0]));
int N = findClusterIdx(cl_0, clusters);
sb.append("C " + N + " "); // \n");
}
if (((int[]) cluster_idxs.get(idxs[1])).length == 1)
sb.append("L " + back_translation[idxs[1]] + " "); // \n\n");
else {
int[] cl_0 = ((int[]) cluster_idxs.get(idxs[1]));
int N = findClusterIdx(cl_0, clusters);
sb.append("C " + N + " "); // \n\n");
}
Vector new_cl = new Vector();
int[] new_arr = appendArrays(cl_1, cl_2);
for (int i = 0; i < new_arr.length; i++) new_cl.add(new Integer(new_arr[i]));
clusters.add(new_cl); // its idx is -1 of the ""official"" idx
cluster_idxs.remove(idxs[1]);
cluster_idxs.set(idxs[0], appendArrays(cl_1, cl_2));
// mngmt op:
for (int j = idxs[1]; j < cluster_idxs.size(); j++)
back_translation[j] = back_translation[j + 1];
for (int j = cluster_idxs.size(); j < nr_genes; j++) back_translation[j] = -1;
// 3. recompute the distances
for (int j = 0; j < cluster_idxs.size(); j++) {
switch (linkage_method) {
case MethodConstants.SINGLE_LINKAGE:
((Vector) cluster_dists.get(idxs[0]))
.set(
j,
min(
(Double) ((Vector) cluster_dists.get(idxs[0])).get(j),
(Double) ((Vector) cluster_dists.get(idxs[1])).get(j)));
break;
case MethodConstants.COMPLETE_LINKAGE:
((Vector) cluster_dists.get(idxs[0]))
.set(
j,
max(
(Double) ((Vector) cluster_dists.get(idxs[0])).get(j),
(Double) ((Vector) cluster_dists.get(idxs[1])).get(j)));
break;
case MethodConstants.AVERAGE_LINKAGE:
((Vector) cluster_dists.get(idxs[0]))
.set(
j,
avg(
(Double) ((Vector) cluster_dists.get(idxs[0])).get(j),
(Double) ((Vector) cluster_dists.get(idxs[1])).get(j),
cl_1.length,
cl_2.length));
break;
default:
break; /* not foreseen! */
}
}
for (int i = 0; i <= cluster_idxs.size(); i++)
((Vector) cluster_dists.get(i)).remove(idxs[1]);
cluster_dists.remove(idxs[1]);
for (int i = 0; i < cluster_idxs.size(); i++) {
if (i != idxs[0])
((Vector) cluster_dists.get(i))
.set(idxs[0], ((Vector) cluster_dists.get(idxs[0])).get(i));
}
// repeat 2 and 3 until all items are in a single cluster
}
HCL_output = sb.toString();
if (MethodConstants.debug)
System.out.println("nr_items, nr_joins: " + nr_genes + ", " + nr_joins);
System.out.println("Here and GUI is : " + owner + ", and engine is : " + engine);
int[] chipsTemplate = new int[engine.getPre().getWorkingChipCount()];
System.out.println("lenght of array is : " + chipsTemplate.length);
for (int i = 0; i < chipsTemplate.length; i++) chipsTemplate[i] = i;
// ....
quasi_bi_clusters = new LinkedList();
for (int i = 0; i < cluster_idxs.size(); i++) {
int[] genes = (int[]) cluster_idxs.get(i);
int[] chips = chipsTemplate.clone();
bicat.biclustering.Bicluster bc = new bicat.biclustering.Bicluster(i + 1, genes, chips);
quasi_bi_clusters.add(bc);
}
if (engine == null) System.out.println("Engine is null!");
engine.setData(engine.getPre().getPreprocessedData());
if (owner != null) {
owner.getPp().setData(engine.getRawData());
int[] temp = (int[]) cluster_idxs.get(0);
bicat.biclustering.Bicluster selection =
new bicat.biclustering.Bicluster(0, temp.clone(), chipsTemplate);
engine.setCurrentBiclusterSelection(owner.getPp().setTranslationTable(selection));
owner.getMatrixScrollPane().repaint();
owner.readjustPictureSize();
owner.getPp().repaint();
}
}