updated extreme_muts.R with number of budding hotspots and mult muts numbers

scripts/plotting/extreme_muts.R

@@ -124,3 +124,40 @@ hs_foldx
 #===============
 
 aa_muts = merged_df3[merged_df3$ligand_distance<5,]
+
+aa_dist = paste0(5, angstroms_symbol)
+
+cat("No. of active site residues within", aa_dist, ":", nrow(aa_muts))
+
+#====================
+# budding hotspots
+#====================
+
+# Method 
+# this is what you want
+foo = merged_df3 %>% group_by(position) %>% tally()
+bar = merged_df3 %>% group_by(position) %>% count()
+
+# sanity check
+all(table(foo$n) == table(bar$n))
+table(foo$n)
+
+
+n_budding_sites = table(foo$n)[[2]]
+n_mult_muts_sites = sum(table(foo$n)) - (table(foo$n)[[1]] - table(foo$n)[[2]])
+
+cat("No of budding hotspots (sites with 2 mutations):", n_budding_sites
+    , "\nNo. of sites with mutiple (>2) mutations:", n_mult_muts_sites)
+
+# another way
+setDT(merged_df3)[, pos_count := .N, by = .(position)]
+
+# this is cummulative
+table(merged_df3$pos_count)
+
+# use group by on this: same as the 
+snpsBYpos_df <- merged_df3 %>%
+  group_by(position) %>%
+  summarize(snpsBYpos = mean(pos_count))
+
+