getPopulationFit() error after XDE test

[riddhee-mehta]

Hello, after doing an XDE test, when I try to run getPopulationFit(), I get the following error:
Res_all$populationFit <- getPopulationFit(Res_all, gene = diffgene_all, type = 'Variable', num.timepoint = 1000) Error in dimnames(x) <- dn : length of 'dimnames' [2] not equal to array extent

Traceback:

What could be the cause for this error? My data has ~26000 cells.

Thanks!

[Winnie09]

Hi Mehta,

It seems that there is a mismatch between the colnames of your gene expression matrix and the pseudotime names. Both should be cell names. We've seen similar errors when: e.g. no names for the pseudotime (missing cell names), the gene expression matrix has column names with "cell.1" but the pseudotime names are "cell_1", etc.. Please try and let me know if you have further questions!

Best,
Wenpin

[riddhee-mehta]

Hi Wenpin,

I checked the following and am still getting the same error:
`> length(colnames(Res_all$expr)) [1] 26830

length(names(Res_all$pseudotime)) [1] 26830
sum(colnames(Res_all$expr) %in% names(Res_all$pseudotime)) [1] 26830`

Please let me know if there is anything else I can check. Thanks!

[daniel-spies]

I've encountered tow bugs in the population fit function in the last code block that return either only cell names or potentially duplicated cell names lines 109-122 failing for the xde and tde test subsequent plotting functions

if (type == 'VARIABLE'){
    fitres <- lapply(names(fitlist[[1]]), function(i){
      tmp <- t(sapply(fitlist, function(j){
        j[[i]]
      }))
      colnames(tmp) = names(pseudotime)[pt]
    })  
    names(fitres) <- names(fitlist[[1]])
  } 
} else if (type == 'TIME'){
    fitres <- t(do.call(cbind, fitlist))
    rownames(fitres) <- gene
    if (ncol(testobj$expr) == ncol(fitres)) colnames(fitres) <- colnames(testobj$expr)
    colnames(fitres) = names(pseudotime)[pt]
  }

should be changed to the following:

1. return the tmp df not the colnames
2. there should be an else statement as the colnames are otherwise anyways overwritten

if (type == 'VARIABLE'){
    fitres <- lapply(names(fitlist[[1]]), function(i){
      tmp <- t(sapply(fitlist, function(j){
        j[[i]]
      }))
      colnames(tmp) = names(pseudotime)[pt]
      return(tmp)
    })  
    names(fitres) <- names(fitlist[[1]])
  } 
} else if (type == 'TIME'){
    fitres <- t(do.call(cbind, fitlist))
    rownames(fitres) <- gene
    if (ncol(testobj$expr) == ncol(fitres)){
        colnames(fitres) <- colnames(testobj$expr)
    } else {
        colnames(fitres) = names(pseudotime)[pt]
    }
  }