Merge branch 'master' of gitbio.ens-lyon.fr:LBMC/hub/hic

benchmark/plots_contact.r

+library(tidyverse)
+library(LaF)
+#######################################
+#                                     #
+#   PLOTS ON NUMBER OF CONTACTS       #
+#                                     #
+#######################################
+
+path_root = "/Users/laurent/projects/physbio/hic/results/"
+
+count_contact <- function(x) {
+  read_log(x, col_names = F)$X5[1]
+}
+
+data <- tibble(
+  file = paste0(path_root, "/", list.files(path = path_root, pattern =".*matrix_sparse\\.log", recursive = T))
+) %>%
+  mutate(
+    name = str_extract(file, ".*\\/(.*_.*_[^\\/]*)\\/", group = 1),
+    correction = map(name, function(x){str_split(x, "_")[[1]][2]}),
+    counts = map(file, count_contact)
+  ) %>%
+  unnest(c(correction, counts)) %>%
+  mutate(
+    algorithm = ifelse(str_detect(correction, ".*cutsite.*"), "cutsite", ifelse(str_detect(correction, ".*parasplit.*"), "parasplit", NA)),
+    option = ifelse(str_detect(correction, ".*fr.*"), "forward-backward", ifelse(str_detect(correction, ".*cutsite.*"), "forward-backward", "all")),
+    option = ifelse(str_detect(correction, ".*seed0.*"), paste0(option, ", seed=0"), paste0(option, ", seed=20")),
+    name = paste(algorithm, option)
+  )
+
+
+save(data, file = "number_of_contacts.Rdata")
+
+
+ggplot(data, aes(x = reorder(name,counts), y = counts, color = algorithm, shape = option)) +
+  geom_point(size = 5) +
+  scale_y_log10() +
+  theme_bw() +
+  theme(axis.title.x=element_blank(),
+        axis.text.x=element_blank(),
+        axis.ticks.x=element_blank()) +
+  ggtitle("Number of contacts")
+ggsave("number_of_contacts.pdf")
+