private static void initializeIndex() {
indexProvider = new LuceneBatchInserterIndexProvider(inserter);
nodeIndex = indexProvider.nodeIndex("nodeIndex", MapUtil.stringMap("type", "exact"));
// TODO: Does this have an effect at all?
nodeIndex.setCacheCapacity(NodeKeys.TYPE, 100000);
nodeIndex.setCacheCapacity(NodeKeys.NAME, 100000);
nodeIndex.setCacheCapacity(NodeKeys.CODE, 100000);
}
@Ignore
@Test
public void testInsertionSpeed() {
BatchInserter inserter = BatchInserters.inserter(new File(PATH, "3").getAbsolutePath());
BatchInserterIndexProvider provider = new LuceneBatchInserterIndexProvider(inserter);
BatchInserterIndex index = provider.nodeIndex("yeah", EXACT_CONFIG);
index.setCacheCapacity("key", 1000000);
long t = currentTimeMillis();
for (int i = 0; i < 1000000; i++) {
Map<String, Object> properties = map("key", "value" + i);
long id = inserter.createNode(properties);
index.add(id, properties);
}
System.out.println("insert:" + (currentTimeMillis() - t));
index.flush();
t = currentTimeMillis();
for (int i = 0; i < 1000000; i++) {
count((Iterator<Long>) index.get("key", "value" + i));
}
System.out.println("get:" + (currentTimeMillis() - t));
}
public static void main(String[] args) {
if (args.length != 3) {
System.out.println(
"This program expects the following parameters: \n"
+ "1. Folder name with all the .gbk files \n"
+ "2. Bio4j DB folder \n"
+ "3. batch inserter .properties file");
} else {
File currentFolder = new File(args[0]);
File[] files = currentFolder.listFiles();
BatchInserter inserter = null;
BatchInserterIndexProvider indexProvider = null;
// ----------------------------------------------------------------------------------
// ---------------------initializing node type properties----------------------------
genomeElementProperties.put(
GenomeElementNode.NODE_TYPE_PROPERTY, GenomeElementNode.NODE_TYPE);
geneProperties.put(GeneNode.NODE_TYPE_PROPERTY, GeneNode.NODE_TYPE);
cdsProperties.put(CDSNode.NODE_TYPE_PROPERTY, CDSNode.NODE_TYPE);
miscRnaProperties.put(MiscRNANode.NODE_TYPE_PROPERTY, MiscRNANode.NODE_TYPE);
mRnaProperties.put(MRNANode.NODE_TYPE_PROPERTY, MRNANode.NODE_TYPE);
ncRnaProperties.put(NcRNANode.NODE_TYPE_PROPERTY, NcRNANode.NODE_TYPE);
rRnaProperties.put(RRNANode.NODE_TYPE_PROPERTY, RRNANode.NODE_TYPE);
tmRnaProperties.put(TmRNANode.NODE_TYPE_PROPERTY, TmRNANode.NODE_TYPE);
tRnaProperties.put(TRNANode.NODE_TYPE_PROPERTY, TRNANode.NODE_TYPE);
// ----------------------------------------------------------------------------------
// ----------------------------------------------------------------------------------
try {
// This block configures the logger with handler and formatter
fh = new FileHandler("ImportRefSeq.log", false);
SimpleFormatter formatter = new SimpleFormatter();
fh.setFormatter(formatter);
logger.addHandler(fh);
logger.setLevel(Level.ALL);
// create the batch inserter
inserter = BatchInserters.inserter(args[1], MapUtil.load(new File(args[2])));
// create the batch index service
indexProvider = new LuceneBatchInserterIndexProvider(inserter);
// -----------------create batch indexes----------------------------------
// ----------------------------------------------------------------------
BatchInserterIndex genomeElementVersionIndex =
indexProvider.nodeIndex(
GenomeElementNode.GENOME_ELEMENT_VERSION_INDEX,
MapUtil.stringMap(PROVIDER_ST, LUCENE_ST, TYPE_ST, EXACT_ST));
BatchInserterIndex nodeTypeIndex =
indexProvider.nodeIndex(
Bio4jManager.NODE_TYPE_INDEX_NAME,
MapUtil.stringMap(PROVIDER_ST, LUCENE_ST, TYPE_ST, EXACT_ST));
for (File file : files) {
if (file.getName().endsWith(".gbff")) {
logger.log(Level.INFO, ("file: " + file.getName()));
BufferedReader reader = new BufferedReader(new FileReader(file));
String line = null;
while ((line = reader.readLine()) != null) {
// this is the first line where the locus is
String accessionSt = "";
String definitionSt = "";
String versionSt = "";
String commentSt = "";
StringBuilder seqStBuilder = new StringBuilder();
ArrayList<String> cdsList = new ArrayList<String>();
ArrayList<String> geneList = new ArrayList<String>();
ArrayList<String> miscRnaList = new ArrayList<String>();
ArrayList<String> mRnaList = new ArrayList<String>();
ArrayList<String> ncRnaList = new ArrayList<String>();
ArrayList<String> rRnaList = new ArrayList<String>();
ArrayList<String> tmRnaList = new ArrayList<String>();
ArrayList<String> tRnaList = new ArrayList<String>();
boolean originFound = false;
// Now I get all the lines till I reach the string '//'
do {
boolean readLineFlag = true;
if (line.startsWith(GBCommon.LOCUS_STR)) {
// do nothing right now
} else if (line.startsWith(GBCommon.ACCESSION_STR)) {
accessionSt = line.split(GBCommon.ACCESSION_STR)[1].trim();
} else if (line.startsWith(GBCommon.VERSION_STR)) {
versionSt = line.split(GBCommon.VERSION_STR)[1].trim().split(" ")[0];
} else if (line.startsWith(GBCommon.DEFINITION_STR)) {
definitionSt += line.split(GBCommon.DEFINITION_STR)[1].trim();
do {
line = reader.readLine();
if (line.startsWith(" ")) {
definitionSt += line.trim();
}
} while (line.startsWith(" "));
readLineFlag = false;
} else if (line.startsWith(GBCommon.COMMENT_STR)) {
commentSt += line.split(GBCommon.COMMENT_STR)[1].trim();
do {
line = reader.readLine();
if (line.startsWith(" ")) {
commentSt += "\n" + line.trim();
}
} while (line.startsWith(" "));
readLineFlag = false;
} else if (line.startsWith(GBCommon.FEATURES_STR)) {
do {
line = reader.readLine();
String lineSubstr5 = line.substring(5);
if (lineSubstr5.startsWith(GBCommon.CDS_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.CDS_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
cdsList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.GENE_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.GENE_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
geneList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.MISC_RNA_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.MISC_RNA_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
miscRnaList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.TM_RNA_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.TM_RNA_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
tmRnaList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.R_RNA_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.R_RNA_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
rRnaList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.M_RNA_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.M_RNA_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
mRnaList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.NC_RNA_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.NC_RNA_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
ncRnaList.add(positionsSt);
} else if (lineSubstr5.startsWith(GBCommon.T_RNA_STR)) {
String positionsSt = "";
positionsSt += line.trim().split(GBCommon.T_RNA_STR)[1].trim();
line = reader.readLine();
while (!line.trim().startsWith("/")) {
positionsSt += line.trim();
line = reader.readLine();
}
tRnaList.add(positionsSt);
}
} while (line.startsWith(" "));
readLineFlag = false;
} else if (line.startsWith(GBCommon.ORIGIN_STR)) {
originFound = true;
do {
line = reader.readLine();
String[] tempArray = line.trim().split(" ");
for (int i = 1; i < tempArray.length; i++) {
seqStBuilder.append(tempArray[i]);
}
} while (line.startsWith(" "));
readLineFlag = false;
}
if (readLineFlag) {
line = reader.readLine();
}
} while (line != null && !line.startsWith(GBCommon.LAST_LINE_STR));
// --------create genome element node--------------
long genomeElementId =
createGenomeElementNode(
versionSt,
commentSt,
definitionSt,
inserter,
genomeElementVersionIndex,
nodeTypeIndex);
// -----------genes-----------------
for (String genePositionsSt : geneList) {
geneProperties.put(GeneNode.POSITIONS_PROPERTY, genePositionsSt);
long geneId = inserter.createNode(geneProperties);
inserter.createRelationship(genomeElementId, geneId, genomeElementGeneRel, null);
// indexing gene node by its node_type
nodeTypeIndex.add(
geneId, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, GeneNode.NODE_TYPE));
}
// -----------CDS-----------------
for (String cdsPositionsSt : cdsList) {
cdsProperties.put(CDSNode.POSITIONS_PROPERTY, cdsPositionsSt);
long cdsID = inserter.createNode(cdsProperties);
inserter.createRelationship(genomeElementId, cdsID, genomeElementCDSRel, null);
// indexing CDS node by its node_type
nodeTypeIndex.add(
cdsID, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, CDSNode.NODE_TYPE));
}
// -----------misc rna-----------------
for (String miscRnaPositionsSt : miscRnaList) {
miscRnaProperties.put(MiscRNANode.POSITIONS_PROPERTY, miscRnaPositionsSt);
long miscRnaID = inserter.createNode(miscRnaProperties);
inserter.createRelationship(
genomeElementId, miscRnaID, genomeElementMiscRnaRel, null);
// indexing MiscRNA node by its node_type
nodeTypeIndex.add(
miscRnaID,
MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, MiscRNANode.NODE_TYPE));
}
// -----------m rna-----------------
for (String mRnaPositionsSt : mRnaList) {
mRnaProperties.put(MRNANode.POSITIONS_PROPERTY, mRnaPositionsSt);
long mRnaID = inserter.createNode(mRnaProperties);
inserter.createRelationship(genomeElementId, mRnaID, genomeElementMRnaRel, null);
// indexing MRNA node by its node_type
nodeTypeIndex.add(
mRnaID, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, MRNANode.NODE_TYPE));
}
// -----------nc rna-----------------
for (String ncRnaPositionsSt : ncRnaList) {
ncRnaProperties.put(NcRNANode.POSITIONS_PROPERTY, ncRnaPositionsSt);
long ncRnaID = inserter.createNode(ncRnaProperties);
inserter.createRelationship(genomeElementId, ncRnaID, genomeElementNcRnaRel, null);
// indexing NCRNA node by its node_type
nodeTypeIndex.add(
ncRnaID, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, NcRNANode.NODE_TYPE));
}
// -----------r rna-----------------
for (String rRnaPositionsSt : rRnaList) {
rRnaProperties.put(RRNANode.POSITIONS_PROPERTY, rRnaPositionsSt);
long rRnaID = inserter.createNode(rRnaProperties);
inserter.createRelationship(genomeElementId, rRnaID, genomeElementRRnaRel, null);
// indexing RRNA node by its node_type
nodeTypeIndex.add(
rRnaID, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, RRNANode.NODE_TYPE));
}
// -----------tm rna-----------------
for (String tmRnaPositionsSt : tmRnaList) {
tmRnaProperties.put(TmRNANode.POSITIONS_PROPERTY, tmRnaPositionsSt);
long tmRnaID = inserter.createNode(tmRnaProperties);
inserter.createRelationship(genomeElementId, tmRnaID, genomeElementTmRnaRel, null);
// indexing TmRNA node by its node_type
nodeTypeIndex.add(
tmRnaID, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, TmRNANode.NODE_TYPE));
}
// -----------t rna-----------------
for (String tRnaPositionsSt : tRnaList) {
tRnaProperties.put(TRNANode.POSITIONS_PROPERTY, tRnaPositionsSt);
long tRnaID = inserter.createNode(tRnaProperties);
inserter.createRelationship(genomeElementId, tRnaID, genomeElementTRnaRel, null);
// indexing TRNA node by its node_type
nodeTypeIndex.add(
tRnaID, MapUtil.map(Bio4jManager.NODE_TYPE_INDEX_NAME, TRNANode.NODE_TYPE));
}
logger.log(Level.INFO, (versionSt + " saved!"));
}
}
}
} catch (Exception e) {
logger.log(Level.SEVERE, e.getMessage());
StackTraceElement[] trace = e.getStackTrace();
for (StackTraceElement stackTraceElement : trace) {
logger.log(Level.SEVERE, stackTraceElement.toString());
}
} finally {
// shutdown, makes sure all changes are written to disk
indexProvider.shutdown();
inserter.shutdown();
// closing logger file handler
fh.close();
}
}
}