Empirical Bayes Change-point Model

  • "Essentially, all models are wrong, but some are useful---George E. P. Box"
  • "Essentially, all models are wrong, but some are useful---George E. P. Box "

Usage:

perl   EBChangePoint.pl   -c   < control.bam >   -t   < case.bam >   -g   <exon.bed >   -h1   < control_junction >   -h2   < case_junction >   [options]

Required:

-c   FILE   Control sample in bam format.

-t   FILE   Case sample in bam format.

-g   FILE   Regions to be analyzed in bed format, such as exon regions.

-h1   FILE   Junction.bed file for control sample, i.e. generated by Tophat.

-h2   FILE   Junction.bed file for case sample, i.e. generated by Tophat.

Options:

-o   STRING   Output file. [change_point]

-a   FLOAT   Mixed directional FDR level. [0.05]

-s   STRING   Max heap size for JVM. [256m]

-x   STRING   Min heap size for JVM. [1024m]

Example:

perl   path/to/EBChangePoint.pl   -c   control.bam   -t   case.bam   -g   exon.bed   -h1   junctions_control.bed   -h2   junctions_case.bed

Understand Output:

COLUMN MEANING
Chr chromosome name
Start start position
End end postion
Exon Exon name
isAS final decision based on specified FDR level. (0 for no; 1 for yes)
CP if isAS=1, CP records the estimated change-point position
Control_Max_Prob the posterior probability for most likely change-point position for this exon in control sample
Control_NCP_Prob the posterior probability for no change-point for this exon in control sample
Case_Max_Prob the posterior probability for most likely change-point position for this exon in case sample
Case_NCP_Prob the posterior probability for no change-point for this exon in case sample