/** Test of processNextStart method, of class Overlap. */
@Test
public void testProcessNextStart() throws IOException {
System.out.println("Tabix Test!");
// Example direct tabix query and the results... (note only the JSON will be returned by the
// overlapPipe)
// r0240560:tabix m102417$ tabix genes.tsv.bgz 17:41196312-41300000
// 17 41196312 41277500
// {"_type":"gene","_landmark":"17","_strand":"-","_minBP":41196312,"_maxBP":41277500,"gene":"BRCA1","gene_synonym":"BRCAI; BRCC1; BROVCA1; IRIS; PNCA4; PPP1R53; PSCP; RNF53","note":"breast cancer 1, early onset; Derived by automated computational analysis using gene prediction method: BestRefseq.","GeneID":"672","HGNC":"1100","HPRD":"00218","MIM":"113705"}
// 17 41231278 41231833
// {"_type":"gene","_landmark":"17","_strand":"+","_minBP":41231278,"_maxBP":41231833,"gene":"RPL21P4","gene_synonym":"RPL21_58_1548","note":"ribosomal protein L21 pseudogene 4; Derived by automated computational analysis using gene prediction method: Curated Genomic.","pseudo":"","GeneID":"140660","HGNC":"17959"}
// 17 41277600 41297130
// {"_type":"gene","_landmark":"17","_strand":"+","_minBP":41277600,"_maxBP":41297130,"gene":"NBR2","gene_synonym":"NCRNA00192","note":"neighbor of BRCA1 gene 2 (non-protein coding); Derived by automated computational analysis using gene prediction method: BestRefseq.","GeneID":"10230","HGNC":"20691"}
// 17 41286808 41287385
// {"_type":"gene","_landmark":"17","_strand":"+","_minBP":41286808,"_maxBP":41287385,"gene":"LOC100505873","note":"Derived by automated computational analysis using gene prediction method: GNOMON. Supporting evidence includes similarity to: 1 EST, 1 Protein","pseudo":"","GeneID":"100505873"}
// 17 41296973 41297272
// {"_type":"gene","_landmark":"17","_strand":"+","_minBP":41296973,"_maxBP":41297272,"gene":"HMGN1P29","note":"high mobility group nucleosome binding domain 1 pseudogene 29; Derived by automated computational analysis using gene prediction method: Curated Genomic.","pseudo":"","GeneID":"100885865","HGNC":"39373"}
// String query =
// "my\tfirst\tquery\t{\"_landmark\":\"17\",\"_minBP\":41196312,\"_maxBP\":41277500}"; //1
// result
String query2 =
"my\tfirst\tquery\t{\"_landmark\":\"17\",\"_minBP\":41196312,\"_maxBP\":41300000}"; // 5
// results
List<History> result = new ArrayList<History>();
OverlapPipe op = new OverlapPipe(geneFile);
Pipe<String, History> p2 = new Pipeline<String, History>(new HistoryInPipe(), op);
p2.setStarts(Arrays.asList(query2));
while (p2.hasNext()) {
result.add(p2.next());
}
assertEquals(5, result.size()); // 5 rows returned
// System.out.println(result.get(4)); //5 rows returned
}
@Test
public void testProcessNextStart_OneResult() throws IOException {
System.out.println("Tabix Test.. TWO RESULTS!");
// Example direct tabix query and the results... (note only the JSON will be returned by the
// overlapPipe)
// r0240560:tabix m102417$ tabix genes.tsv.bgz 17:41196312-41300000
// 17 41196312 41277500
// {"_type":"gene","_landmark":"17","_strand":"-","_minBP":41196312,"_maxBP":41277500,"gene":"BRCA1","gene_synonym":"BRCAI; BRCC1; BROVCA1; IRIS; PNCA4; PPP1R53; PSCP; RNF53","note":"breast cancer 1, early onset; Derived by automated computational analysis using gene prediction method: BestRefseq.","GeneID":"672","HGNC":"1100","HPRD":"00218","MIM":"113705"}
// {"_type":"gene","_landmark":"17","_strand":"+","_minBP":41231278,"_maxBP":41231833,"gene":"RPL21P4","gene_synonym":"RPL21_58_1548","note":"ribosomal protein L21 pseudogene 4; Derived by automated computational analysis using gene prediction method: Curated Genomic.","pseudo":"","GeneID":"140660","HGNC":"17959"}
// TODO find a query that returns exactly 1 row
String query =
"my\tfirst\tquery\t{\"_landmark\":\"17\",\"_minBP\":41196312,\"_maxBP\":41277500}"; // 2
// results
List<History> result = new ArrayList<History>();
OverlapPipe op = new OverlapPipe(geneFile);
Pipe<String, History> p2 = new Pipeline<String, History>(new HistoryInPipe(), op);
p2.setStarts(Arrays.asList(query));
while (p2.hasNext()) {
result.add(p2.next());
}
// System.out.println("size="+result.size());
assertEquals(2, result.size()); // two result
}
@Test
public void testBRCA1DOWN() throws IOException {
System.out.println("Get the genes down-chromosome and including BRCA1");
String[] paths = {"gene"};
String query = "x\ty\tz\t" + BRCA1;
OverlapPipe op = new OverlapPipe(geneFile, 0, 19772);
Pipe p = new Pipeline(new HistoryInPipe(), op, new DrillPipe(false, paths), new PrintPipe());
p.setStarts(Arrays.asList(query));
for (int i = 1; p.hasNext(); i++) {
History h = (History) p.next();
if (i == 1) {
assertEquals("BRCA1", h.get(4));
}
if (i == 2) {
assertEquals("RPL21P4", h.get(4)); // gene overlaps BRCA1 so it is included
}
if (i == 3) {
assertEquals("NBR2", h.get(4));
}
if (i == 4) {
assertEquals("LOC100505873", h.get(4));
}
if (i == 5) {
assertEquals("HMGN1P29", h.get(4));
}
}
}
@Test
public void testGetCluster() throws IOException {
System.out.println("Get all the genes around BRCA1");
String[] paths = {"gene"};
String query = "x\ty\tz\t{\"_landmark\":\"17\",\"_minBP\":41166622,\"_maxBP\":41297272}";
OverlapPipe op = new OverlapPipe(geneFile, 0, 0);
Pipe p = new Pipeline(new HistoryInPipe(), op, new DrillPipe(false, paths), new PrintPipe());
p.setStarts(Arrays.asList(query));
for (int i = 1; p.hasNext(); i++) {
History h = (History) p.next();
if (i == 1) {
assertEquals("VAT1", h.get(4));
}
if (i == 2) {
assertEquals("RND2", h.get(4));
}
if (i == 3) {
assertEquals("BRCA1", h.get(4));
}
if (i == 4) {
assertEquals("RPL21P4", h.get(4));
}
if (i == 5) {
assertEquals("NBR2", h.get(4));
}
if (i == 6) {
assertEquals("LOC100505873", h.get(4));
}
if (i == 7) {
assertEquals("HMGN1P29", h.get(4));
}
}
}
public S processNextStart() {
final S s = this.starts.next();
this.pipe.setStarts(new SingleIterator<S>(s));
try {
while (true) {
pipe.next();
}
} catch (final NoSuchElementException e) {
}
return s;
}
// make sure to test zero results!!!
@Test
public void testProcessNextStart_ZeroResults() throws Exception {
System.out.println("Tabix Test.. Zero Results!");
String query =
"my\tfirst\tquery\t{\"_landmark\":\"17\",\"_minBP\":4,\"_maxBP\":41}"; // 1 result
try {
List<History> result = new ArrayList<History>();
OverlapPipe op = new OverlapPipe(geneFile);
Pipe<String, History> p2 = new Pipeline<String, History>(new HistoryInPipe(), op);
p2.setStarts(Arrays.asList(query));
// assertTrue(p.hasNext()==false);
for (int i = 0; p2.hasNext(); i++) {
p2.next();
}
} catch (Exception e) {
Assert.fail("INVALID QUERY... EXPECTED EXCEPTION!!!");
}
}
/**
* 1 49482 rs202079915 G A . .
* RSPOS=49482;dbSNPBuildID=137;SSR=0;SAO=0;VP=050000000005000002000100;WGT=1;VC=SNV;ASP;OTHERK{"CHROM":"1","POS":"49482","ID":"rs202079915","REF":"G","ALT":"A","QUAL":".","FILTER":".","INFO":{"RSPOS":49482,"dbSNPBuildID":137,"SSR":0,"SAO":0,"VP":"050000000005000002000100","WGT":1,"VC":"SNV","ASP":true,"OTHERKG":true},"_id":"rs202079915","_type":"variant","_landmark":"1","_refAllele":"G","_altAlleles":["A"],"_minBP":49482,"_maxBP":49482}
* {}
*/
@Test
public void testProcessNextStart_ZeroResultsEmptyJson() throws Exception {
System.out.println("Tabix Test.. Zero Results.. with Empty JSON!");
String query =
"my\tfirst\tquery\t{\"_landmark\":\"1\",\"_minBP\":49482,\"_maxBP\":49482}"; // 1 result
try {
List<History> result = new ArrayList<History>();
OverlapPipe op = new OverlapPipe(geneFile);
Pipe<String, History> p2 = new Pipeline<String, History>(new HistoryInPipe(), op);
p2.setStarts(Arrays.asList(query));
// assertTrue(p.hasNext()==false);
for (int i = 0; p2.hasNext(); i++) {
result.add(p2.next());
}
// System.out.println(result.size());
assertEquals("{}", result.get(4)); // EMPTY JSON as last element in the list
} catch (Exception e) {
// Assert.fail("INVALID QUERY... EXPECTED EXCEPTION!!!");
}
}
public void testIdEdgePipeGraph() {
Graph graph = TinkerGraphFactory.createTinkerGraph();
List<String> ids = Arrays.asList("9", "11", "12");
Pipe<String, Edge> pipe = new IdEdgePipe<String>(graph);
pipe.setStarts(ids);
int counter = 0;
while (pipe.hasNext()) {
Edge edge = pipe.next();
if (counter == 0) {
assertEquals(edge.getId(), "9");
assertEquals(edge.getProperty("weight"), 0.4f);
} else if (counter == 1) {
assertEquals(edge.getId(), "11");
assertEquals(edge.getInVertex(), graph.getVertex("3"));
} else if (counter == 2) {
assertEquals(edge.getId(), "12");
assertEquals(edge.getLabel(), "created");
} else {
throw new RuntimeException("Illegal state.");
}
counter++;
}
assertEquals(counter, 3);
}
@Test
public void extractSequencePipeline() throws IOException {
System.out.println("Extract Sequence Pipeline Test.");
Pipe p = new Pipeline(new HistoryInPipe(), new Bed2SequencePipe(tabixGenome));
p.setStarts(Arrays.asList(p1, p2, p3, p4, p5));
for (int i = 0; p.hasNext(); i++) {
ArrayList al = (ArrayList) p.next();
if (i == 0) {
assertEquals("G", al.get(3));
}
if (i == 1) {
assertEquals("G", al.get(3));
}
if (i == 2) {
assertEquals("T", al.get(3));
}
if (i == 3) {
assertEquals("C", al.get(3));
}
if (i == 4) {
assertEquals("G", al.get(3));
}
}
}