Made a few improvements to the examine_droplet_logliks.R analysis.

inst/analysis/check_loglik_calcs.R

@@ -1,7 +1,7 @@
 # A script to check our log-likelihood calculations for different
 # packages (glmpca, scGBM, fastglmpca).
 library(Matrix)
-library(fastTopics)
+library(fastglmpca)
 source("../code/loglik.R")
 load("~/git/fastTopics-experiments/data/droplet.RData")
 counts <- t(counts)

inst/analysis/examine_droplet_logliks.R

@@ -0,0 +1,54 @@
+library(Matrix)
+library(fastglmpca)
+library(ggplot2)
+library(cowplot)
+source("../code/loglik.R")
+load("~/git/fastTopics-experiments/data/droplet.RData")
+counts <- t(counts)
+fit1 <- readRDS(paste("droplets_fastglmpca_fit_10_factors_5105_iter_28",
+                      "cores_dec_23.rds",sep = "_"))
+fit2 <- readRDS(paste("droplets_glmpca_fit_10_factors_10_hrs_avagrad",
+                      "optimizer_minibatch_stochastic_dec_23.rds",sep = "_"))
+fit3 <- readRDS("droplets_scGBM_fit_10_factors_no_beta_infer_10_hrs.rds")
+glmpca_pois_loglik <- function (Y, H)
+  Y*H - exp(H) - lfactorial(as.matrix(Y))
+H1 <- glmpca_pois_loglik(counts,logrates_fastglmpca(fit1))
+H2 <- glmpca_pois_loglik(counts,logrates_glmpca(fit2))
+H3 <- glmpca_pois_loglik(counts,logrates_scgbm(fit3))
+
+tissue_colors <- c("royalblue",   # basal
+                   "firebrick",   # ciliated
+                   "forestgreen", # club
+                   "gold",        # goblet
+                   "darkmagenta", # ionocyte
+                   "darkorange",  # neuroendocrine
+                   "skyblue")     # tuft
+
+# Compare cell-wise logliks.
+pdat <- cbind(samples,
+              data.frame(fastglmpca = colSums(H1),
+                         glmpca     = colSums(H2),
+                         scgbm      = colSums(H3)))
+p1 <- ggplot(pdat,aes(x = glmpca,y = fastglmpca,color = tissue)) +
+  geom_point() +
+  geom_abline(intercept = 0,slope = 1,color = "black",linetype = "dotted") +
+  scale_color_manual(values = tissue_colors) +
+  theme_cowplot(font_size = 12)
+p2 <- ggplot(pdat,aes(x = scgbm,y = fastglmpca,color = tissue)) +
+  geom_point() +
+  geom_abline(intercept = 0,slope = 1,color = "black",linetype = "dotted") +
+  scale_color_manual(values = tissue_colors) +
+  theme_cowplot(font_size = 12)
+
+# Compare gene-wise logliks.
+pdat <- data.frame(fastglmpca = rowSums(H1),
+                   glmpca     = rowSums(H2),
+                   scgbm      = rowSums(H3))
+p3 <- ggplot(pdat,aes(x = glmpca,y = fastglmpca)) +
+  geom_point() +
+  geom_abline(intercept = 0,slope = 1,color = "magenta",linetype = "dotted") +
+  theme_cowplot(font_size = 12)
+p4 <- ggplot(pdat,aes(x = scgbm,y = fastglmpca)) +
+  geom_point() +
+  geom_abline(intercept = 0,slope = 1,color = "magenta",linetype = "dotted") +
+  theme_cowplot(font_size = 12)