Sparse to dense conversion even when sparseOptimization is TRUE

[dimalvovs]

sparseOptimization is nice feature that reduces the memory footprint and allows faster computation in CoGAPS. To some reason, sparse matrix is not supported as input, proof:

mat <- Matrix::sparseMatrix(i = sample(1:20, 100, replace = TRUE),
                            j = sample(1:10, 100, replace = TRUE),
                            x = runif(100))

  params <- CogapsParams(seed = 42,
                         nIterations = 30,
                         nPatterns = 2,
                         sparseOptimization = as.logical(0),
                         distributed = "genome-wide")
  params <- setDistributedParams(params, nSets = 2)

  cg <- CoGAPS(mat,
         params = params,
         BPPARAM = BiocParallel::SerialParam(),
         messages = FALSE)

yields:

Error in convertDataToMatrix(data) : unsupported data type

Let's add native sparseMatrix support.

Thanks to Brian@IGS for heads up!