Matt:LabNotes/2015-4-15

From ZhangLabWiki
Jump to navigation Jump to search

Rank Correlation Between DARTFISH and FISSEQ[edit]

  • Repeat these linear regressions with rank correlation to see if it can improve p-value
    • Since relationship between DARTFISH and FISSEQ is not expected to be linear, rank correlation may be more appropriate
  • Workspace: "C:\Users\Matt\Dropbox\GradZhangLab\DecodeData\ReferenceData\.RData"

Spearman's Correlation[edit]

 V4V7_GeneCounts$RankFISSEQv1 <- rank(V4V7_GeneCounts$FISSEQv1, na.last="keep",ties.method="average")
 V4V7_GeneCounts$RankFISSEQv2 <- rank(V4V7_GeneCounts$FISSEQv2, na.last="keep",ties.method="average")
 V4V7_GeneCounts$RankFibV4_Norm <- rank(V4V7_GeneCounts$FibV4_Norm, na.last="keep",ties.method="average")
  • FISSEQv1 vs FISSEQv2
 > cor.test(V4V7_GeneCounts$RankFISSEQv1, V4V7_GeneCounts$RankFISSEQv2, method="spearman")
   Spearman's rank correlation rho
 data:  V4V7_GeneCounts$RankFISSEQv1 and V4V7_GeneCounts$RankFISSEQv2
 S = 2198.517, p-value = 3.244e-08
 alternative hypothesis: true rho is not equal to 0
 sample estimates:
     rho 
 0.7594357 
  • FISSEQv1 vs DARTFISH
 > cor.test(V4V7_GeneCounts$RankFISSEQv1, V4V7_GeneCounts$RankFibV4_Norm, method="spearman")
   Spearman's rank correlation rho
 data:  V4V7_GeneCounts$RankFISSEQv1 and V4V7_GeneCounts$RankFibV4_Norm
 S = 867037.8, p-value = 0.216
 alternative hypothesis: true rho is not equal to 0
 sample estimates:
      rho 
 0.09292391 
  • FISSEQv2 vs DARTFISH
 > cor.test(V4V7_GeneCounts$RankFibV4_Norm, V4V7_GeneCounts$RankFISSEQv2, method="spearman")
   Spearman's rank correlation rho
 data:  V4V7_GeneCounts$RankFibV4_Norm and V4V7_GeneCounts$RankFISSEQv2
 S = 5811.43, p-value = 0.02462
 alternative hypothesis: true rho is not equal to 0
 sample estimates:
     rho 
 0.3641066
  • Could potentially improve correlation by not ranking genes in DARTFISH that are not present in FISSEQ...