In zeropin/TFCookbook: Analyzing protein-DNA interactions data in principled way

require(dplyr)
require(stringi)

reference = "CCACTAGGGGGCGCTA"

ZFP57

ZFP57 %>%
  group_by(Property) %>%
  summarise(n())

ZFP57 <- ZFP57 %>%
  dplyr::mutate(`Bound/Unbound` = Bound/Unbound,
                Energy = -log(Bound/Unbound))

ZFP57 %>%
  dplyr::filter(Property == "un") %>%
  TFCookbook::buildEnergyModel(encoding = "3L+1") %>%
  TFCookbook::getEnergyMatrix(encoding = "3L+1") %>%
  TFCookbook::plotEnergyLogo()

require(ggplot2)
require(ggrepel)

CTCF.un <- subset(CTCF, Property=="un")
CTCF.me <- subset(CTCF, Property=="me")

inner_join(CTCF.un, CTCF.me, by = "Sequence") %>%
  dplyr::select(Sequence,
         Energy.un = `Energy.x`,
         Energy.me = `Energy.y`) %>%
  dplyr::mutate(CpG.containing = as.factor(stringi::stri_count_fixed(Sequence, "CG")),
                Label = if_else(abs(Energy.un - Energy.me)>1, Sequence, "")) %>%
  ggplot(aes(x = Energy.un, y = Energy.me, shape = CpG.containing, color = CpG.containing)) +
  geom_label_repel(aes(label = Label), show.legend = FALSE) +
  geom_point() +
  geom_abline(intercept = -0.25, slope = 1, linetype="dashed") +
  geom_abline(intercept = 0.25, slope = 1, linetype="dashed") +
  ggtitle("Binding energy comparison between M.SssI treated and untreated sites") +
  xlab("Binding energy for untreated sites (kT)") + ylab("Binding energy for M.SssI treated sites (kT)")

ZFP57 %>%
  dplyr::mutate(Sequence = if_else(Property=="M.SssI",
                                   stringi::stri_replace_all_fixed(Sequence, "CG", "MW"),
                                   Sequence)) %>%
  dplyr::mutate(Sequence = if_else(Property=="dM",
                                   stringi::stri_replace_all_fixed(Sequence, "TGCCGC", "TGCMWC"),
                                   Sequence)) %>%
  dplyr::mutate(Sequence = if_else(Property=="tH",
                                   stringi::stri_replace_all_fixed(Sequence, "TGCCGC", "TGCMGC"),
                                   Sequence)) %>%
  dplyr::mutate(Sequence = if_else(Property=="bH",
                                   stringi::stri_replace_all_fixed(Sequence, "TGCCGC", "TGCCWC"),
                                   Sequence)) %>%
  TFCookbook::buildEnergyModel(encoding = "5L+1") %>%
  TFCookbook::getEnergyMatrix(encoding = "5L+1") %>%
  TFCookbook::plotEnergyLogo()