update one-sided fisher test to two-sided

core/src/main/java/org/mskcc/cbio/portal/servlet/CalcFisherExactTest.java

@@ -44,13 +44,13 @@
 import org.mskcc.cbio.portal.stats.FisherExact;
 
 /**
- * Calculate the cumulative (one-tail) p-value out of fisher exact test
+ * Calculate the cumulative (two-tail) p-value out of fisher exact test
  * 
  * @param a a, b, c, d are the four cells in a 2x2 matrix
  * @param b
  * @param c
  * @param d
- * @return one-tailed P-value (right or left, whichever is smallest)
+ * @return two-tailed P-value (right or left, whichever is smallest)
  * 
  */
 public class CalcFisherExactTest extends HttpServlet {
@@ -86,7 +86,7 @@ protected void doPost(HttpServletRequest httpServletRequest,
  int c = Integer.parseInt(dataSet.split(" ")[2]);
  int d = Integer.parseInt(dataSet.split(" ")[3]); 
  FisherExact fisher = new FisherExact(a + b + c + d);
- double pValue = fisher.getCumlativeP(a, b, c, d);
+ double pValue = fisher.getTwoTailedP(a, b, c, d);
  result = result.concat(String.valueOf(pValue) + " "); 
  }
  }

core/src/main/java/org/mskcc/cbio/portal/stats/OddsRatio.java

@@ -87,7 +87,7 @@ public OddsRatio(ProfileDataSummary pDataSummary, String geneA, String geneB) {
 
  oddsRatio = ((double) (a * d)) / ((double) (b * c));
  FisherExact fisher = new FisherExact(a + b + c + d);
- pValue = fisher.getCumlativeP(a, b, c, d);
+ pValue = fisher.getTwoTailedP(a, b, c, d);
  lowerConfidenceInterval = Math.exp(Math.log(oddsRatio) - 1.96 * (Math.sqrt(1 / (double) a
  + 1 / (double) b + 1 / (double) c + 1 / (double) d)));
  upperConfidenceInterval = Math.exp(Math.log(oddsRatio) + 1.96 * (Math.sqrt(1 / (double) a
@@ -103,7 +103,7 @@ public double getOddsRatio() {
  return oddsRatio;
  }
 
- public double getCumulativeP() {
+ public double getTwoTailedP() {
  return pValue;
  }
 

core/src/main/java/org/mskcc/cbio/portal/util/EnrichmentsAnalysisUtil.java

@@ -652,7 +652,7 @@ private double runFisherExactTest(HashMap<Integer, String> singleGeneCaseValueMa
  }
  }
  FisherExact fisher = new FisherExact(a + b + c + d);
- return fisher.getCumlativeP(a, b, c, d);
+ return fisher.getTwoTailedP(a, b, c, d);
  }
 
  private String getCytoband(int geneticEntityId, String geneticProfileStableId) {

core/src/test/java/org/mskcc/cbio/portal/stats/TestFisherExact.java

@@ -39,7 +39,7 @@ public class TestFisherExact extends TestCase {
 
  public void testFisherExact() {
  FisherExact fisher = new FisherExact(43 + 2 + 17 + 7);
- double p = fisher.getCumlativeP(43, 2, 17, 7);
+ double p = fisher.getTwoTailedP(43, 2, 17, 7);
  assertEquals(0.006653, p, 0.00001);
  }
 }

core/src/test/java/org/mskcc/cbio/portal/stats/TestOddsRatio.java

@@ -70,9 +70,9 @@ public void testFisherExact() throws IOException {
 
  OddsRatio oddsRatio = new OddsRatio(pDataSummary, "BRCA1", "BRCA2");
  double oddsRatioValue = oddsRatio.getOddsRatio();
- double p = oddsRatio.getCumulativeP();
+ double p = oddsRatio.getTwoTailedP();
  assertEquals(0.16666, oddsRatioValue, 0.0001);
- assertEquals(0.0849, p, 0.0001);
+ assertEquals(0.1698, p, 0.0001);
  //System.out.println (oddsRatio.getRCommand());
  }
 }

docs/user-guide/faq.md

@@ -146,7 +146,7 @@ Group Comparison is a suite of analysis features which allows a user to compare
 - Other tabs
  - 2 groups
  - Continuous data: one-sided t-test
- - Categorical data: one-sided Fisher's exact test
+ - Categorical data: two-sided Fisher's exact test
  - 3 or more groups
  - Continuous data: one-way ANOVA
  - Categorical data: Chi-squared test

persistence/persistence-mybatis/src/test/java/org/cbioportal/persistence/mybatis/AlterationMyBatisRepositoryTest.java

@@ -1034,7 +1034,7 @@ public void getPatientStructuralVariantCountIncludeCustomDriverAnnotationsInclud
  alterationFilter.setStructuralVariants(true);
  alterationFilter.setMutationTypeSelect(Select.none());
  alterationFilter.setCnaTypeSelect(Select.none());
- alterationFilter.setSelectedTiers(Select.byValues(List.of("Class 2")));
+ alterationFilter.setSelectedTiers(Select.byValues(Arrays.asList("Class 2")));
  alterationFilter.setIncludeUnknownTier(true);
  List<AlterationCountByGene> result = alterationMyBatisRepository.getPatientAlterationGeneCounts(
  svPatientIdToProfileId,
@@ -1049,7 +1049,7 @@ public void getPatientStructuralVariantCountIncludeCustomDriverAnnotationsExclud
  alterationFilter.setStructuralVariants(true);
  alterationFilter.setMutationTypeSelect(Select.none());
  alterationFilter.setCnaTypeSelect(Select.none());
- alterationFilter.setSelectedTiers(Select.byValues(List.of("Class 2")));
+ alterationFilter.setSelectedTiers(Select.byValues(Arrays.asList("Class 2")));
  alterationFilter.setIncludeUnknownTier(false);
  List<AlterationCountByGene> result = alterationMyBatisRepository.getPatientAlterationGeneCounts(
  svPatientIdToProfileId,

service/src/main/java/org/cbioportal/service/util/AlterationEnrichmentUtil.java

@@ -132,7 +132,7 @@ public List<AlterationEnrichment> createAlterationEnrichments(Map<String, Pair<L
  int alteredOnlyInQueryGenesCount = counts.get(0).getProfiledCount()
  - counts.get(0).getAlteredCount();
 
- pValue = fisherExactTestCalculator.getCumulativePValue(alteredInNoneCount,
+ pValue = fisherExactTestCalculator.getTwoTailedPValue(alteredInNoneCount,
  counts.get(1).getAlteredCount(), alteredOnlyInQueryGenesCount,
  counts.get(0).getAlteredCount());
  } else {

service/src/main/java/org/cbioportal/service/util/FisherExactTestCalculator.java

@@ -40,4 +40,51 @@ public double getCumulativePValue(int a, int b, int c, int d) {
  }
  return p;
  }
+
+ public double getTwoTailedPValue(int a, int b, int c, int d) {
+
+ int min, i;
+ int n = a + b + c + d;
+ double p = 0;
+ double[] f = new double[n + 1];
+ f[0] = 0.0;
+
+ for (int j = 1; j <= n; j++) {
+ f[j] = f[j - 1] + Math.log(j);
+ }
+
+ double baseP = getPValue(a, b, c, d, f);
+// in order for a table under consideration to have its p-value included
+// in the final result, it must have a p-value less than the baseP, i.e.
+// Fisher's exact test computes the probability, given the observed marginal
+// frequencies, of obtaining exactly the frequencies observed and any configuration more extreme.
+// By "more extreme," we mean any configuration (given observed marginals) with a smaller probability of
+// occurrence in the same direction (one-tailed) or in both directions (two-tailed).
+
+ int initialA = a, initialB = b, initialC = c, initialD = d;
+ p += baseP;
+
+ min = (c < b) ? c : b;
+ for (i = 0; i < min; i++) {
+ double tempP = getPValue(++a, --b, --c, ++d, f);
+ if (tempP <= baseP) {
+ p += tempP;
+ }
+ }
+
+ // reset the values to their original so we can repeat this process for the other side
+ a = initialA;
+ b = initialB;
+ c = initialC;
+ d = initialD;
+
+ min = (a < d) ? a : d;
+ for (i = 0; i < min; i++) {
+ double pTemp = getPValue(--a, ++b, ++c, --d, f);
+ if (pTemp <= baseP) {
+ p += pTemp;
+ }
+ }
+ return p;
+ }
 }

service/src/test/java/org/cbioportal/service/util/AlterationEnrichmentUtilTest.java

@@ -101,8 +101,8 @@ public void createAlterationEnrichments() throws Exception {
 
  // START: for 2 groups
 
- Mockito.when(fisherExactTestCalculator.getCumulativePValue(1, 1, 2, 0)).thenReturn(1.0);
- Mockito.when(fisherExactTestCalculator.getCumulativePValue(2, 0, 0, 2)).thenReturn(0.3);
+ Mockito.when(fisherExactTestCalculator.getTwoTailedPValue(1, 1, 2, 0)).thenReturn(1.0);
+ Mockito.when(fisherExactTestCalculator.getTwoTailedPValue(2, 0, 0, 2)).thenReturn(0.3);
 
  List<AlterationEnrichment> result = alterationEnrichmentUtil.createAlterationEnrichments(
  mutationCountsbyEntrezGeneIdAndGroup);