/** To compare motif found by each methods with TF.jtux using STAMP */
private static void compareTFStampResults(String[] args) {
final int TOP_MOTIF_RANK = 9;
double stamp_p_value = Double.parseDouble(args[3]);
// load the mapping file between tf and known motif db entries
// String[] lines = readSmallTextFile(args[0]);
// HashMap<String, HashSet<String>> tf2db = new HashMap<String, HashSet<String>>();
// for (int i=0;i<lines.length;i++){
// String[] fs = lines[i].split("\t");
// if (fs.length<=1)
// continue;
// String tf = fs[0];
// HashSet<String> entries = new HashSet<String>();
// for (int j=1;j<fs.length;j++){
// String entry = fs[j].trim();
// if (entry.length()>=1)
// entries.add(entry);
// }
// tf2db.put(tf, entries);
// }
// load encode expts, and motif methods
String[] expts = readSmallTextFile(args[0]); // read expt/tf pairs
HashMap<String, String> expt2tf = new HashMap<String, String>();
for (int i = 0; i < expts.length; i++) {
String[] fs = expts[i].split("\t");
expt2tf.put(fs[0].trim(), fs[1].trim());
expts[i] = fs[0].trim(); // replace with expt only
}
String[] methods = readSmallTextFile(args[1]);
// load STAMP file for each expt_method pair
HashMap<String, Integer> performances = new HashMap<String, Integer>();
File stamp_result_dir = new File(args[2]);
for (String expt : expts) {
// String tf = expt2tf.get(expt);
// if (tf2db.containsKey(tf)){
each_method:
for (String method : methods) {
String expt_method = expt + "." + method;
File f = new File(stamp_result_dir, expt_method + "_match_pairs.txt");
if (!f.exists()) continue;
String[] sls = readSmallTextFile(f.getAbsolutePath()); // stampe lines
for (int i = 0; i < sls.length - 1; i += 2) {
if (sls[i].startsWith(">")) {
int rank = i / 2; // motif rank in this expt
// only check the top match to TF.jtux
String[] sl_fs = sls[i + 1].split("\t");
// String entry = sl_fs[0].trim();
double p = Double.parseDouble(sl_fs[1]);
if (p < stamp_p_value) {
performances.put(expt_method, rank);
continue each_method;
}
}
}
}
// }
// else {
// System.err.println(tf + " is not in the TF_2_DB list: " + args[0]);
// }
}
// print out results
StringBuilder sb = new StringBuilder("expt\ttf\t");
for (String method : methods) {
sb.append(method).append("\t");
}
sb.append("\n");
HashMap<String, Integer> tf2count = new HashMap<String, Integer>();
for (String expt : expts) {
String tf = expt2tf.get(expt);
// if (!tf2db.containsKey(tf)) // only count annotated expts/tfs with public known
// motif
// continue;
if (tf2count.containsKey(tf)) tf2count.put(tf, tf2count.get(tf) + 1);
else tf2count.put(tf, 1);
sb.append(expt).append("\t").append(tf).append("\t");
for (String method : methods) {
String pair = expt + "." + method;
int rank = performances.containsKey(pair) ? performances.get(pair) : 99;
sb.append(rank).append("\t");
}
sb.append("\n");
}
CommonUtils.writeFile("method_rank_matrix.txt", sb.toString());
// print out result for each top rank
HashMap<String, int[]> performanceByExpt = new HashMap<String, int[]>();
HashMap<String, float[]> performanceByTF = new HashMap<String, float[]>();
sb = new StringBuilder("Rank\t");
StringBuilder sb2 = new StringBuilder("Rank\t");
for (String method : methods) {
sb.append(method).append("\t");
sb2.append(method).append("\t");
performanceByExpt.put(method, new int[9]);
performanceByTF.put(method, new float[9]);
}
sb.append("\n");
sb2.append("\n");
for (String expt : expts) {
String tf = expt2tf.get(expt);
for (String method : methods) {
String pair = expt + "." + method;
int rank = performances.containsKey(pair) ? performances.get(pair) : 99;
for (int r = 0; r < TOP_MOTIF_RANK; r++) {
if (rank <= r) {
performanceByExpt.get(method)[r]++;
performanceByTF.get(method)[r] += 1.0 / tf2count.get(tf);
}
}
}
}
for (int r = 0; r < TOP_MOTIF_RANK; r++) {
sb.append("Top" + r + "\t");
sb2.append("Top" + r + "\t");
for (String method : methods) {
int[] scores = performanceByExpt.get(method);
float[] scores_tf = performanceByTF.get(method);
sb.append(scores[r]).append("\t");
sb2.append(String.format("%.2f\t", scores_tf[r]));
}
sb.append("\n");
sb2.append("\n");
}
CommonUtils.writeFile("method_expt_scores.txt", sb.toString());
CommonUtils.writeFile("method_tf_scores.txt", sb2.toString());
}
/** To compare motif found by each methods with known motif in database using STAMP */
private static void compareStampResults(String[] args) {
final int STAMP_UNIT_LINE_COUNT = 11;
int stamp_top_count = Integer.parseInt(args[5]);
double stamp_p_value = Double.parseDouble(args[6]);
// load the mapping file between tf and known motif db entries
String[] lines = readSmallTextFile(args[0]);
HashMap<String, HashSet<String>> tf2db = new HashMap<String, HashSet<String>>();
for (int i = 1; i < lines.length; i++) { // skip line 0, header
String[] fs = lines[i].split("\t");
if (fs.length <= 1) continue;
String tf = fs[0];
HashSet<String> entries = new HashSet<String>();
for (int j = 1; j < fs.length; j++) {
String entry = fs[j].trim();
if (entry.length() >= 1) entries.add(entry);
}
tf2db.put(tf, entries);
}
// load motif clusters
lines = readSmallTextFile(args[1]);
ArrayList<HashSet<String>> clusters = new ArrayList<HashSet<String>>();
for (int i = 0; i < lines.length; i++) {
String[] fs = lines[i].split("Cluster_Members:\t");
if (fs.length <= 1) continue;
String[] ms = fs[1].split("\t");
HashSet<String> cluster = new HashSet<String>();
for (String m : ms) {
m = m.trim();
if (m.length() >= 1) cluster.add(m);
}
clusters.add(cluster);
}
// expand tf to db entry mapping
for (String tf : tf2db.keySet()) {
HashSet<String> expandedSet = new HashSet<String>();
HashSet<String> annotations = tf2db.get(tf);
for (String anno : annotations) {
for (HashSet<String> cluster : clusters) {
if (cluster.contains(anno)) expandedSet.addAll(cluster);
}
}
tf2db.put(tf, expandedSet);
}
// load encode expts, and motif methods
String[] expts = readSmallTextFile(args[2]);
HashMap<String, String> expt2tf = new HashMap<String, String>();
for (int i = 0; i < expts.length; i++) {
String[] fs = expts[i].split("\t");
expt2tf.put(fs[0].trim(), fs[1].trim());
expts[i] = fs[0].trim();
}
String[] methods = readSmallTextFile(args[3]);
// load STAMP file for each expt_method pair
HashMap<String, Integer> performances = new HashMap<String, Integer>();
File dir = new File(args[4]);
for (String expt : expts) {
String tf = expt2tf.get(expt);
if (tf2db.containsKey(tf)) {
each_method:
for (String method : methods) {
String pair = expt + "." + method;
File f = new File(dir, pair + "_match_pairs.txt");
if (!f.exists()) continue;
String[] sls = readSmallTextFile(f.getAbsolutePath()); // stampe lines
for (int i = 0; i < sls.length; i += STAMP_UNIT_LINE_COUNT) {
if (sls[i].startsWith(">")) {
int rank = i / STAMP_UNIT_LINE_COUNT; // motif rank in this expt
for (int j = 1;
j <= stamp_top_count;
j++) { // each top db entry in STAMP match results
String[] sl_fs = sls[i + j].split("\t");
String entry = sl_fs[0].trim();
if (tf2db.get(tf).contains(entry)) {
double p = Double.parseDouble(sl_fs[1]);
if (p < stamp_p_value) {
performances.put(pair, rank);
continue each_method;
}
}
}
}
}
}
}
}
// print out results
StringBuilder sb = new StringBuilder("expt\ttf\t");
for (String method : methods) {
sb.append(method).append("\t");
}
sb.append("\n");
HashMap<String, Integer> tf2count = new HashMap<String, Integer>();
for (String expt : expts) {
String tf = expt2tf.get(expt);
if (!tf2db.containsKey(tf)) // only count annotated expts/tfs with public known motif
continue;
if (tf2count.containsKey(tf)) tf2count.put(tf, tf2count.get(tf) + 1);
else tf2count.put(tf, 1);
sb.append(expt).append("\t").append(tf).append("\t");
for (String method : methods) {
String pair = expt + "." + method;
int rank = performances.containsKey(pair) ? performances.get(pair) : 99;
sb.append(rank).append("\t");
}
sb.append("\n");
}
CommonUtils.writeFile("method_rank_matrix.txt", sb.toString());
// print out result for each top rank
HashMap<String, int[]> performanceByExpt = new HashMap<String, int[]>();
HashMap<String, float[]> performanceByTF = new HashMap<String, float[]>();
sb = new StringBuilder("Rank\t");
StringBuilder sb2 = new StringBuilder("Rank\t");
for (String method : methods) {
sb.append(method).append("\t");
sb2.append(method).append("\t");
performanceByExpt.put(method, new int[9]);
performanceByTF.put(method, new float[9]);
}
sb.append("\n");
sb2.append("\n");
for (String expt : expts) {
String tf = expt2tf.get(expt);
for (String method : methods) {
String pair = expt + "." + method;
int rank = performances.containsKey(pair) ? performances.get(pair) : 99;
for (int r = 0; r < 9; r++) {
if (rank <= r) {
performanceByExpt.get(method)[r]++;
performanceByTF.get(method)[r] += 1.0 / tf2count.get(tf);
}
}
}
}
for (int r = 0; r < 9; r++) {
sb.append("Top" + r + "\t");
sb2.append("Top" + r + "\t");
for (String method : methods) {
int[] scores = performanceByExpt.get(method);
float[] scores_tf = performanceByTF.get(method);
sb.append(scores[r]).append("\t");
sb2.append(String.format("%.2f\t", scores_tf[r]));
}
sb.append("\n");
sb2.append("\n");
}
CommonUtils.writeFile("method_expt_scores.txt", sb.toString());
CommonUtils.writeFile("method_tf_scores.txt", sb2.toString());
}