private void increment(List<String> features, int transition, float delta) {
for (String feat : features) {
if (!data.containsKey(feat)) {
data.put(feat, new FeatureWeights());
}
data.get(feat).increment(transition, delta, i);
}
}
public float weight(Collection<String> features, int trans) {
float score = 0.0f;
for (String feature : features) {
if (data.containsKey(feature)) {
score += data.get(feature).scoreForTransition(trans);
}
}
return score;
}
public float[] weight(Collection<String> features) {
float[] scores = new float[index.size()];
for (String feature : features) {
if (!data.containsKey(feature)) {
continue;
}
if (isAveraged || data.get(feature).used > opts.minupdate) {
data.get(feature).sumScoresForAllTransitions(scores);
}
}
return scores;
}
/**
* Return all Genbank IDs associated to this probeset.
*
* @param probesetID
* @return Genbank IDs associated to this probeset.
*/
public String[] getGenbankList(final String probesetID) {
final String gbList = (String) probesetId2GenbankList.get(probesetID);
if (gbList == null) {
return ArrayUtils.EMPTY_STRING_ARRAY;
} else {
return gbList.split("[ ]");
}
}
public Weights<T> average() {
System.err.print("averaging (this may take a while)... ");
Weights<T> result = new Weights<T>(opts, true);
result.data = new Object2ObjectOpenHashMap<>();
result.index = index;
result.gram = gram;
int cnt = 0;
for (String feat : data.keySet()) {
FeatureWeights dw = data.get(feat);
if (dw.used > opts.minupdate) {
cnt++;
FeatureWeights fw = new FeatureWeights();
for (int trans : index.indices()) {
float averaged = dw.getAveraged(trans, i);
if (!Float.isNaN(averaged)) {
fw.increment(trans, averaged, 0);
}
}
result.data.put(feat, fw);
}
}
System.err.println("done, averaged " + cnt + " features.");
return result;
}
/**
* Utility method to check whether a given walker has been deprecated in a previous GATK release
*
* @param walkerName the walker class name (not the full package) to check
*/
public static boolean isDeprecatedWalker(final String walkerName) {
return deprecatedGATKWalkers.containsKey(walkerName);
}
static {
// Same comments as for walkers
deprecatedGATKAnnotations.put("DepthOfCoverage", "2.4 (renamed to Coverage)");
}
static {
// Indicate recommended replacement in parentheses if applicable
deprecatedGATKWalkers.put(
"ReduceReads", "3.0 (use recommended best practices pipeline with the HaplotypeCaller)");
deprecatedGATKWalkers.put(
"CountCovariates", "2.0 (use BaseRecalibrator instead; see documentation for usage)");
deprecatedGATKWalkers.put(
"TableRecalibration",
"2.0 (use PrintReads with -BQSR instead; see documentation for usage)");
deprecatedGATKWalkers.put("AlignmentWalker", "2.2 (no replacement)");
deprecatedGATKWalkers.put("CountBestAlignments", "2.2 (no replacement)");
deprecatedGATKWalkers.put(
"SomaticIndelDetector", "2.0 (replaced by MuTect2; see documentation for usage)");
deprecatedGATKWalkers.put(
"BeagleOutputToVCF",
"3,4 (replaced by Beagle native functions; see Beagle 4 documentation at https://faculty.washington.edu/browning/beagle/beagle.html)");
deprecatedGATKWalkers.put(
"VariantsToBeagleUnphased",
"3.4 (replaced by Beagle native functions; see Beagle 4 documentation at https://faculty.washington.edu/browning/beagle/beagle.html)");
deprecatedGATKWalkers.put(
"ProduceBeagleInput",
"3.4 (replaced by Beagle native functions; see Beagle 4 documentation at https://faculty.washington.edu/browning/beagle/beagle.html)");
deprecatedGATKWalkers.put(
"ReadAdaptorTrimmer",
"3.5 (this tool was unsound and untested -- no specific replacement, see Picard tools for alternatives)");
deprecatedGATKWalkers.put(
"BaseCoverageDistribution",
"3.5 (use DiagnoseTargets instead; see documentation for usage)");
deprecatedGATKWalkers.put(
"CoveredByNSamplesSites", "3.5 (use DiagnoseTargets instead; see documentation for usage)");
deprecatedGATKWalkers.put(
"VariantValidationAssessor", "3.5 (this tool was unsound and untested -- no replacement)");
deprecatedGATKWalkers.put(
"LiftOverVariants", "3.5 (use Picard LiftoverVCF instead; see documentation for usage)");
deprecatedGATKWalkers.put(
"FilterLiftedVariants",
"3.5 (use Picard LiftoverVCF instead; see documentation for usage)");
deprecatedGATKWalkers.put(
"ListAnnotations", "3.5 (this tool was impractical; see the online documentation instead)");
}
/**
* Utility method to pull up the version number at which an annotation was deprecated and the
* suggested replacement, if any
*
* @param annotationName the annotation class name (not the full package) to check
*/
public static String getAnnotationDeprecationInfo(final String annotationName) {
return deprecatedGATKAnnotations.get(annotationName).toString();
}
/**
* Utility method to pull up the version number at which a walker was deprecated and the suggested
* replacement, if any
*
* @param walkerName the walker class name (not the full package) to check
*/
public static String getWalkerDeprecationInfo(final String walkerName) {
return deprecatedGATKWalkers.get(walkerName).toString();
}
/**
* Utility method to check whether a given annotation has been deprecated in a previous GATK
* release
*
* @param annotationName the annotation class name (not the full package) to check
*/
public static boolean isDeprecatedAnnotation(final String annotationName) {
return deprecatedGATKAnnotations.containsKey(annotationName);
}
/**
* Read the platform information from a GEO platform file.
*
* @param filename
* @throws SyntaxErrorException
* @throws IOException
*/
public void read(final String filename) throws SyntaxErrorException, IOException {
if (filename == null) {
return;
}
final GeoPlatformFileReader reader = new GeoPlatformFileReader();
final Reader lowLevelReader;
if (filename.endsWith(".gz")) {
lowLevelReader = new InputStreamReader(new GZIPInputStream(new FileInputStream(filename)));
} else {
lowLevelReader = new FileReader(filename);
}
final Table platformFileContent = reader.read(lowLevelReader);
final Table.RowIterator ri = platformFileContent.firstRow();
int probeSetIdColumnIndex = -1;
int genbankAcColumnIndex = -1;
int genbankListColumnIndex = -1;
int genbankColumnIndex = -1;
final String probeIdColumnName = "ID";
final String genbankAcColumnName = "GB_ACC";
final String genbankListColumnName = "GB_LIST";
try {
probeSetIdColumnIndex = platformFileContent.getColumnIndex(probeIdColumnName);
} catch (InvalidColumnException e) {
assert false;
}
try {
genbankAcColumnIndex = platformFileContent.getColumnIndex(genbankAcColumnName);
} catch (InvalidColumnException e) { // NOPMD
// OK, see below.
}
try {
genbankListColumnIndex = platformFileContent.getColumnIndex(genbankListColumnName);
} catch (InvalidColumnException e) { // NOPMD
// OK, see below.
}
genbankColumnIndex =
(genbankListColumnIndex != -1 ? genbankListColumnIndex : genbankAcColumnIndex);
if (probeSetIdColumnIndex == -1 || genbankColumnIndex == -1) {
throw new SyntaxErrorException(
0,
"One of the following column names could not be found in the platformFileContent description file: "
+ probeIdColumnName
+ ", (at least one of : "
+ genbankAcColumnName
+ ", "
+ genbankListColumnName
+ " ).");
}
int count = 0;
while (!ri.end()) {
try {
final String probesetId = (String) platformFileContent.getValue(probeSetIdColumnIndex, ri);
final String genbankAccession =
(String) platformFileContent.getValue(genbankColumnIndex, ri);
probesetId2GenbankList.put(probesetId, genbankAccession);
count++;
} catch (TypeMismatchException e) {
throw new InternalError("Column type does not match" + e.getMessage());
} catch (InvalidColumnException e) {
throw new InternalError("Should never happen");
}
ri.next();
}
this.count = count;
}
