Camoco icon indicating copy to clipboard operation
Camoco copied to clipboard

GWAS GO-Simulations non constant candidates

Open schae234 opened this issue 8 years ago • 0 comments

As FCR increases, there are random genes within the GO term that are not within the COB that get replaced with genes that ARE in the COB which results in an increase in candidate genes list as FCR goes up.

offending code in SimulateGWAS.py :

        if num_genes == 0:                                                      
            continue                                                            
        eloci = term.effective_loci(                                            
            window_size=window_size                                             
        )                                                                       
        if args.percent_fcr != None and args.percent_fcr > 0:                   
            # replace some loci with random genes if FDR specified              
            num_fcr = int(len(eloci) * args.percent_fcr)                        
            fcr_loci = cob.refgen.random_genes(num_fcr,window=window_size)      
            # permute and truncate the loci then add fcr loci                   
            fcr_loci = np.concatenate([                                         
                np.random.permutation(eloci)[0:-1*len(fcr_loci)],               
                np.array(list(fcr_loci))                                        
            ])                                                                  
            assert len(fcr_loci) == len(eloci)                                  
            eloci = fcr_loci                                                    
        candidates = cob.refgen.candidate_genes(                                
            eloci,                                                              
            flank_limit=flank_limit                                             
        )

TODO: fix and write a regression test

schae234 avatar Mar 30 '16 15:03 schae234