/** Get info string for a specific allele */
public String getInfo(String key, String allele) {
if (info == null) parseInfo();
// Get INFO value
String infoStr = info.get(key);
if (infoStr == null) return null;
// Split INFO value and match it to allele
String infos[] = infoStr.split(",");
// INFO fields having number type 'R' (all alleles) should have one value for reference as well.
// So in those cases we must skip the first value
int firstAltIndex = 0;
VcfHeaderInfo vcfInfo = getVcfInfo(key);
if (vcfInfo != null && vcfInfo.isNumberAllAlleles()) {
firstAltIndex = 1;
// Are we looking for 'REF' information?
if (ref.equalsIgnoreCase(allele)) return infos[0];
}
// Find ALT matching allele
for (int i = 0, j = firstAltIndex; (i < alts.length) && (j < infos.length); i++, j++)
if (alts[i].equalsIgnoreCase(allele)) return infos[j];
return null;
}
/** Check info field Note: We report the first error we find */
String checkInfo(String infoName) {
if (infoName.isEmpty()) return "";
VcfHeaderInfo vcfInfo = getVcfInfo(infoName);
if (vcfInfo == null) return "Cannot find header for INFO field '" + infoName + "'";
// Split INFO value and match it to allele
String valsStr = getInfo(infoName);
if (valsStr == null) return ""; // INFO field not present, nothing to do
// Check values
String values[] = valsStr.split(",");
for (String val : values)
if (!VcfEntry.isValidInfoValue(val))
return "INFO filed '"
+ infoName
+ "' has an invalid value '"
+ val
+ "' (no spaces, tabs, '=' or ';' are allowed)";
// Check number of INFO elements
if (vcfInfo.isNumberNumber() && vcfInfo.getNumber() != values.length) {
VcfInfoType type = vcfInfo.getVcfInfoType();
if (type == VcfInfoType.Flag && values.length == 1)
; // OK, flags must have one or zero values
else
return "INFO filed '"
+ infoName
+ "' has 'Number="
+ vcfInfo.getNumber()
+ "' in header, but it contains '"
+ values.length
+ "' elements.";
}
if (vcfInfo.isNumberAllAlleles() && values.length != (alts.length + 1))
return "INFO filed '"
+ infoName
+ "' has 'Number=R' in header, but it contains '"
+ values.length
+ "' elements when there are '"
+ alts.length
+ "' alleles (it should have '"
+ (alts.length + 1)
+ "' elements).";
if (vcfInfo.isNumberAllAlleles() && values.length != alts.length)
return "INFO filed '"
+ infoName
+ "' has 'Number=A' in header, but it contains '"
+ values.length
+ "' elements when there are '"
+ alts.length
+ "' alleles.";
return "";
}
/**
* Add a "key=value" tuple the info field
*
* @param key : INFO key name
* @param value : Can be null if it is a boolean field.
*/
public void addInfo(String key, String value) {
if (!isValidInfoKey(key))
throw new RuntimeException(
"Illegal INFO key / name. Key: \""
+ key
+ "\" does not match regular expression ^[A-Za-z_][0-9A-Za-z_.]*$");
if (!isValidInfoValue(value))
throw new RuntimeException(
"No white-space, semi-colons, or equals-signs are permitted in INFO field values. Name:\""
+ key
+ "\" Value:\""
+ value
+ "\"");
// Remove previous 'key' for INFO field?
removeInfo(key);
// Is this a 'flag'?
boolean isFlag = false;
VcfHeader vcfHeader = vcfFileIterator.getVcfHeader();
if (vcfHeader != null) {
VcfHeaderInfo vcfHeaderInfo = vcfFileIterator.getVcfHeader().getVcfInfo(key);
isFlag = (vcfHeaderInfo != null) && (vcfHeaderInfo.getVcfInfoType() == VcfInfoType.Flag);
}
// Add to info hash (if available)
if (info != null) info.put(key, value);
// Append value to infoStr
String addInfoStr = key + (value != null && !isFlag ? "=" + value : ""); // String to append
if ((infoStr == null) || infoStr.isEmpty()) infoStr = addInfoStr;
else {
if (!infoStr.endsWith(SUB_FIELD_SEP))
infoStr += SUB_FIELD_SEP; // Do we need to add a semicolon?
infoStr += addInfoStr; // Add info string
}
}
/** Get Info number for a given ID */
public VcfInfoType getVcfInfoNumber(String id) {
VcfHeaderInfo vcfInfo = vcfFileIterator.getVcfHeader().getVcfInfo(id);
if (vcfInfo == null) return null;
return vcfInfo.getVcfInfoType();
}
