Open yuyingxie opened 4 years ago
I ran the tutorial code with slight changes and got very weird results. I don't know why some p-values are negative and the meaning of NA.1, NA.2...
data(sce)
sce <- sce[, sce$cluster_id == "B cells"] sce$cluster_id <- droplevels(sce$cluster_id)
gs <- sample(nrow(sce), 100) sce <- sce[gs, ]
res1 <- mmDS(sce, method = "dream",
- n_threads = 2, verbose = FALSE) Warning messages: 1: In DGEList(counts(x), norm.factors = 1/sizeFactors(x)) : norm factors don't multiply to 1 2: In MulticoreParam(n_threads, progressbar = verbose) : MulticoreParam() not supported on Windows, use SnowParam() head(res$
B cells
) gene cluster_id beta SE stat p_val p_adj.loc p_adj.glb 1 CAPG B cells -0.2362395 -0.2753383 -0.736435482 -0.2524438 -0.7573315 -14.38930 2 LAMTOR4 B cells 0.2614172 0.3177080 0.235495272 0.2822484 0.5644968 1.00000 3 DBI B cells -0.9036876 -0.8666395 -3.127177349 -0.8944030 -3.5776119 -67.97463 4 SAMSN1 B cells 0.3661611 0.3861396 0.001764934 0.3711063 0.5644968 1.00000 5 RPS20 B cells -0.2362395 -0.2753383 -0.736435482 -0.2524438 -0.7573315 -14.13685 6 LSP1_ENSG00000130592 B cells 0.2614172 0.3177080 0.235495272 0.2822484 0.5644968 1.00000 ...8 NA.1 NA.2 NA.3 NA.4 NA.5 NA.6 NA.7 1 -0.216052477 -1.5741170322 -0.3096581 -2.317899e-01 2.301275e+00 3.445880e+00 -1.4139107713 3.732890e+00 2 0.077441823 0.4060873201 0.3643984 4.776509e-02 4.667559e-01 5.831000e-01 0.3782685918 7.149070e-01 3 -2.789868165 -3.8763018545 -0.8497789 -4.852705e+00 4.930361e+00 5.909587e+00 -3.7378487185 5.221505e+00 4 0.005272951 0.0001060561 0.3954480 1.217888e-06 8.207761e-07 3.429668e-09 0.0001856016 1.774752e-07 5 -0.216052477 -1.5741170322 -0.3096581 -2.317899e-01 2.301275e+00 3.445880e+00 -1.4139107713 3.732890e+00 6 0.077441823 0.4060873201 0.3643984 4.776509e-02 4.667559e-01 5.831000e-01 0.3782685918 7.149070e-01
My question is that when using the method of nbimon, does the method estimate the scaling factor for each cell? If so, we need to keep at least, say, 2000 genes so that we can have a reasonable estimate for the scaling factor.
Originally posted by @yuyingxie in https://github.com/HelenaLC/muscat/issues/38#issuecomment-675234860