From c5ae81583f19cb75aa2e651d23b9c4e297b19bda Mon Sep 17 00:00:00 2001
From: Adrian Sven Geissler <adrian@rth.dk>
Date: Fri, 6 Nov 2020 15:02:24 +0100
Subject: [PATCH] long UTR adress idea

---
 ideas/long_UTR.R | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)
 create mode 100644 ideas/long_UTR.R

diff --git a/ideas/long_UTR.R b/ideas/long_UTR.R
new file mode 100644
index 00000000..f987e375
--- /dev/null
+++ b/ideas/long_UTR.R
@@ -0,0 +1,40 @@
+source('analysis/00_load.R')
+source('scripts/distance_matching.R')
+
+load('data/01_nicolas.rda')
+load('analysis/02_merging.rda')
+
+
+nicolas$upshifts %>%
+  transmute(id, start = pos, end = pos, strand) %>%
+  distance_matching(
+    bind_rows(
+      select(merging$merged_genes,
+             id = merged_id,
+             start, end, strand),
+      nicolas$all.features %>%
+        filter(is.na(type) | str_detect(type, "3'")) %>%
+        mutate(five = ifelse(strand == '+', start, end)) %>%
+        select(id = locus, start = five, end = five, strand)
+    )
+  ) %>%
+  filter(!antisense) -> cmp
+
+
+cmp %>%
+  group_by(x) %>%
+  summarize_at('distance', min) %>%
+  filter(distance <= 500) %>%
+  ggplot(aes(distance)) +
+  geom_histogram()
+
+cmp %>%
+  filter(distance <= 50) %>%
+  select(x) %>%
+  unique -> foo
+
+cmp %>%
+  anti_join(foo) %>%
+  select(x) %>%
+  unique %>%
+  nrow