New dendrogram tests

3d9234714410b7bb76656c92b6da41429d49b48a ceisenhart Mon Feb 29 10:06:31 2016 -0800 Fixing a bug with the top 10 genes feature, refs #16341 diff --git src/hg/expMatrixToJson/expMatrixToJson.c src/hg/expMatrixToJson/expMatrixToJson.c index bf6a925..19b3938 100644 --- src/hg/expMatrixToJson/expMatrixToJson.c +++ src/hg/expMatrixToJson/expMatrixToJson.c @@ -337,52 +337,61 @@ verbose(3,"Merging...\n"); const struct bioExpVector *kid1 = (const struct bioExpVector *)item1; const struct bioExpVector *kid2 = (const struct bioExpVector *)item2; float kid1Weight = kid1->children / (float)(kid1->children + kid2->children); float kid2Weight = kid2->children / (float)(kid1->children + kid2->children); struct bioExpVector *el; AllocVar(el); AllocArray(el->vector, kid1->count); assert(kid1->count == kid2->count); el->count = kid1->count; el->name = catTwoStrings(kid1->name, kid2->name); int i; int gCount = 0; for (i = 0; i < el->count; ++i) { - el->vector[i] = (kid1Weight*kid1->vector[i] + kid2Weight*kid2->vector[i]); - double diff = abs((kid1Weight*kid1->vector[i] - kid2Weight*kid2->vector[i])); - - if (diff > 0.0){ + if (kid1->vector[i] == kid2->vector[i]) // Were doing thousands of merges, lets cut out useless compute where we can. + { + el->vector[i] = kid1->vector[i]; + continue; + } + el->vector[i] = (kid1Weight*kid1->vector[i] + kid2Weight*kid2->vector[i]); // Weight based on number of children. + float diff; + //if (((float)(kid1Weight*kid1->vector[i])) > ((float)(kid2Weight*kid2->vector[i]))) + // {diff = ((float)(kid1Weight*kid1->vector[i])) - ((float)(kid2Weight*kid2->vector[i]));} + //else {diff = ((float)(kid2Weight*kid2->vector[i])) - ((float)(kid1Weight*kid1->vector[i]));} + if (((float)(kid1->vector[i])) > ((float)(kid2->vector[i]))) + {diff = ((float)(kid1->vector[i])) - ((float)(kid2->vector[i]));} + else {diff = ((float)(kid2->vector[i])) - ((float)(kid1->vector[i]));} ++el->contGenes; ++gCount; int index = i + 1; if (gCount <= 10){ struct slDoubleInt *newGene = slDoubleIntNew(diff, index); slAddHead(&el->topGeneIndeces, newGene); slSort(&el->topGeneIndeces, slDoubleIntCmp); } else{ if (el->vector[i] > el->topGeneIndeces->val){ slPopHead(el->topGeneIndeces); struct slDoubleInt *newGene = slDoubleIntNew(diff, index); slAddHead(&el->topGeneIndeces, newGene); slSort(&el->topGeneIndeces, slDoubleIntCmp); } } } - } + slReverse(&el->topGeneIndeces); el->children = kid1->children + kid2->children; return (struct slList *)(el); } void colorLeaves(struct slRef *leafList) /* Assign colors of rainbow to leaves. */ { float total = 0.0; //double purplePos = 0.80; struct slRef *el, *nextEl; /* Loop through list once to figure out total, since we need to normalize */ for (el = leafList; el != NULL; el = nextEl) { nextEl = el->next; @@ -477,31 +486,31 @@ chopSuffix(pageName); int textSize = 12 - log(nodeCount); //int radius = 540 + 270*log10(nodeCount); int width = 10 * nodeCount; int height = 10 * nodeCount; int labelLength = 10+nameSize*(15-textSize); if (labelLength > 100) labelLength = 100; fprintf(outputFile,"\n"); fprintf(outputFile,"\n"); fprintf(outputFile,"New dendrogram tests\n"); fprintf(outputFile,"\n"); fprintf(outputFile,"\n"); fprintf(outputFile,"\n"); fprintf(outputFile,"\n"); - fprintf(outputFile,"\n"); + fprintf(outputFile,"\n"); fprintf(outputFile,"