*****************************************************************************************************************************************************
                            PARAMETRIC ESTIMATOR OF THE LOCAL FDR
*****************************************************************************************************************************************************

DOWNLOAD:   
--------      
            
  The parametric local FDR estimator code is downloadable from the website http://www.people.vcu.edu/~jbukszar/

  The file 'lFDRest_new.r' contains all of the codes necessary to run the parametric local FDR estimator. 



HOW TO USE THE CODE?
--------------------

 The syntax of the code is 


     resu <- localFDR_estimator(inp_x,first,MeinRice=TRUE,rule=1,delrange=c(0, 20))

 where
     
     THE OUTPUT 

        resu:      4 by 'first' array (see first in the inputs), whose
                     first row is the local FDR estimates in increasing order
                     second row is indices of test statistic values 'inp_x' corresponding to the local FDR esimates
                     third row is the test statistic values 'inp_x' corresponding to the local FDR esimates
                     forth row is the estimated number of positive effects repeated 'first' times.

                   

     THE INPUTS
 
         inp_x:     an array containing the test statistic values (which are assumed to follow chi-square d.f. 1 (e.g. Pearson's test stat value 
                    from allele-based case-control studies)   

         first:     the number of the (highest) tests statistic values whose Local FDR's are estimated (if 'first' is negative, then
                    the Local FDR estimate to every test statistic value is calculated)     

         MeinRic:   if MeinRic is TRUE then Meinshausen-Rice method is used to estimate the number of positive effects, 
                    if MeinRic is FALSE then Conservative method is used to estimate the number of positive effects   
     
         rule:      the number added to the estimate of the number of positive effects in the stopping rule (should be either 1 or 0)

         delrange:  the range in which the effect sizes are searched


HOW TO TEST THE CODE?
--------------------

 The code tests the local FDR estimator on simulated data.


 The syntax of the code is 

     lfdrest <- tester_lfdr(m,delr,N,blocksiz=5,blockval=0,rule=1,delrange=c(0, 20))

where 

     THE OUTPUT 

        lfdrest:      N by K array, where K=2*number of positive detectabilities and see N in the inputs. 
                     Each row of array 'lfdrest' corresponds to one simulated dataset and contains the local FDR estimates 
                     of the first K markers/hypotheses.

                 
     THE INPUTS

          m:         number of markers/hypotheses

          delr:      array of positive detectabilities

          N:         number of datasets simulated

          blocksiz:  number of the markers/hypotheses in a group whose test statistics are correlated

          blockval:  the correlation between the test statistics of the markers/hypotheses in a group
     
          rule:      the number added to the estimate of the number of positive effects in the stopping rule (should be either 1 or 0)

          delrange:  the range in which the effect sizes are searched