Thiolutin_Isoniazid.Rmd

---
title: "Supplementary analyses/inspections for thiolutin/isoniazid"
output: html_notebook
editor_options: 
  chunk_output_type: console
---

```{r}
source("./setup.R")
library(readxl)
```

# Get data

```{r}
indir <- "/Volumes/typas/Florian/dbsetup_tables_new"
(nichols_2011 <- read_delim(file.path(indir, "nichols_2011.csv"), delim = ";"))
# same for Lucia's data:
(newsize <- read_delim(file.path(indir, "newsize.csv"), delim = ";"))

# map to new names
ugi_to_name <- read_delim("./data/All_genes_of_E._coli_K-12_substr._MG1655__ECOCYC_to_GENENAME.txt", 
  delim = "\t")
colnames(ugi_to_name) <- c("gene_synonym_new", "ugi")
nichols_2011 <- left_join(nichols_2011, ugi_to_name)
newsize <- left_join(newsize, ugi_to_name)
```


# Thiolutin

Is part of Lucia's data.

## Chemical genetics

Get genetic interactions with cell wall machinery components.

```{r}
cw_comps <- c("mepA", "mepS", "mepH", "mepK", "mepM", "mrcA", "mrcB", "lpoA", 
  "lpoB", "dacB", "pbpG", "amiA", "amiB", "amiC", "ampD", "dacA", "envC", 
  "mpaA", "ddpX", "ycbK")

cw_comps %in% newsize$gene_synonym_new

filter(newsize, gene_synonym_new %in% cw_comps & drugname_typaslab == "THIOLUTIN") %>% 
  select(drugname_typaslab, gene_synonym_new, conc, sscore, qvalue, gene_synonym, 
    ugi, strain) %>% 
  mutate(signif = qvalue <= 0.05) %>% 
  arrange(gene_synonym_new)
```

## Morphological phenotypes

Get quantified phenotypes from Campos..Jacobs-Wagner.

```{r}
campos <- read_xlsx("./data/Other_studies/Campos_Jacobs-Wagner_2018/msb177573-sup-0004-datasetev2.xlsx", 
  sheet = 4, range = "A1:AG4228", na = c("", "NaN"))

my_names <- c("gene", "plt_num", "well_num", "length", "cv_length", "width", 
  "cv_width", "area", "cv_area", "vol", "cv_vol", "sa", "cv_sa", "perim", 
  "cv_perim")
colnames(campos)[1:length(my_names)] <- my_names
# the wt cells
campos_ctrls <- read_xlsx("./data/Other_studies/Campos_Jacobs-Wagner_2018/msb177573-sup-0004-datasetev2.xlsx", 
  sheet = 4, range = "A4229:AG4468", col_names = colnames(campos), 
  na = c("", "NaN"))
campos_ctrls$plt_num <- parse_number(campos_ctrls$plt_num)

# scores for significances
campos_scores <- read_xlsx("./data/Other_studies/Campos_Jacobs-Wagner_2018/msb177573-sup-0004-datasetev2.xlsx", 
  sheet = 2, range = "A1:AG4228", na = c("", "NaN"))

colnames(campos_scores)[1:length(my_names)] <- my_names
# the wt cells
campos_ctrls_scores <- read_xlsx("./data/Other_studies/Campos_Jacobs-Wagner_2018/msb177573-sup-0004-datasetev2.xlsx", 
  sheet = 2, range = "A4229:AG4468", col_names = colnames(campos_scores), 
  na = c("", "NaN"))
campos_ctrls_scores$plt_num <- parse_number(campos_ctrls_scores$plt_num)
```

Get an idea of wild type and mutant sizes. 

```{r}
# vars of interest
voi <- c("gene", "length", "cv_length", "width", "cv_width", "area", "cv_area", 
  "perim", "cv_perim")

newtab <- bind_rows(filter(campos, gene %in% cw_comps), campos_ctrls) %>% 
  select(voi) %>% 
  pivot_longer(-gene, names_to = "param")

# do the same thing for the scores, then join to newtab, flag everything outside 
# of -3 to 3 range
newtab_scores <- bind_rows(filter(campos_scores, gene %in% cw_comps), campos_ctrls_scores) %>% 
  select(voi) %>% 
  pivot_longer(-gene, names_to = "param", values_to = "score")

# ignore wild type stuff
newtab_scores <- filter(newtab_scores, gene %in% cw_comps)

newtab <- left_join(newtab, newtab_scores)
newtab$flag <- newtab$score <= -3 | newtab$score >= 3
newtab$flag[is.na(newtab$flag)] <- FALSE

p <- ggplot(newtab, aes(x = gene, y = value)) + 
  geom_boxplot(aes(colour = flag)) + 
  facet_wrap( ~ param, scales = "free") + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, size = 12))
p

ggsave(filename = "./plots/Morphology_Campos.pdf", plot = p)
```


# Isoniazid

Is part of Nichols data. 

Get all genes with which there is a genetic interaction + the oxidative stress 
response related genes (for peroxides) and the TPP hits. 

```{r}
# genes of interest:
goi <- c("dkgA", "glyA", "katG", "ykgF", "ahpC", "ahpF")

filter(nichols_2011, gene_synonym_new %in% goi & drugname_typaslab == "ISONIAZID") %>% 
  select(drugname_typaslab, gene_synonym_new, conc, sscore, qvalue, gene_synonym, 
    ugi, strain) %>% 
  mutate(signif = qvalue <= 0.05) %>% 
  arrange(gene_synonym_new, conc)

# and now all genes where there is an interaction at qvalue < 0.05
filter(nichols_2011, qvalue < 0.05 & drugname_typaslab == "ISONIAZID") %>% 
  select(drugname_typaslab, gene_synonym_new, conc, sscore, qvalue, gene_synonym, 
    strain) %>% 
  arrange(gene_synonym_new, conc) %>% 
  print(n = 100)
```

See how closely isoniazid resembles triclosan, cerulenin, egcg.

```{r}
my_drugs <- c("ISONIAZID", "TRICLOSAN", "CERULENIN", "EGCG")
# let's pull out each gene that is significant with at least one of the four, 
# then do a correlation plot

interactors <- filter(nichols_2011, drugname_typaslab %in% my_drugs & qvalue < 0.05) %>% 
  pull(gene_synonym) %>% 
  unique()

my_tab <- filter(nichols_2011, drugname_typaslab %in% my_drugs & 
    gene_synonym %in% interactors) %>% 
  select(gene_synonym, drugname_typaslab, conc, sscore)

my_tab$drug_conc <- paste(my_tab$drugname_typaslab, my_tab$conc, sep = "_")
my_tab$drugname_typaslab <- NULL
my_tab$conc <- NULL
my_tab_w <- pivot_wider(my_tab, names_from = drug_conc, values_from = sscore)

library(corrplot)
cor(my_tab_w[, 2:ncol(my_tab_w)], use = "complete.obs", method = "spearman")
```

Stabilisation of DkgA. What about interactions of isoniazid with mutants that lead 
to high levels of methylglyoxal? 

```{r}
# first read in metabolomics table from Campos and Zampieri, 2019
metab_zscores <- read_xlsx("./data/Other_studies/Campos_Zampieri_2019/Zampieri_2019_1-s2.0-S1097276519302710-mmc2.xlsx", 
  sheet = 5, range = "A1:AWH970", na = c("", "NaN"))
metab_pvals <- read_xlsx("./data/Other_studies/Campos_Zampieri_2019/Zampieri_2019_1-s2.0-S1097276519302710-mmc2.xlsx", 
  sheet = 6, range = "A1:AWH970", na = c("", "NaN"))

# consider only zscores <= -3 or >= 3
select(metab_zscores, Metabolite_names, Isoniazid) %>% 
  filter(Isoniazid >= 3 | Isoniazid <= -3) %>% 
  arrange(desc(Isoniazid)) %>% 
  print(n = 15)

# just to check if these metabolites are also upregulated in other drugs 
# but keep only zscores 
select(metab_zscores, Metabolite_names, Isoniazid) %>% 
  filter(Isoniazid >= 3 | Isoniazid <= -3) %>% 
  pull(Metabolite_names) -> 
  iso_metabs

# check p values first:
filter(metab_pvals, Metabolite_names %in% iso_metabs) %>% 
  select(Metabolite_names, Isoniazid) %>% 
  arrange(Isoniazid)

# which other drugs produce these metabolites? focus on z-scores >= 3
select(metab_zscores, 1, 4:last_col()) %>% 
  filter(Metabolite_names %in% iso_metabs) %>% 
  pivot_longer(-Metabolite_names, names_to = "drug", values_to = "zscore") %>% 
  filter(zscore >= 3 | zscore <= -3) %>% 
  arrange(Metabolite_names, desc(zscore)) -> 
  iso_metabs_other_drugs

View(iso_metabs_other_drugs)

# any other drugs that affect all of these metabolites at once?
table(iso_metabs_other_drugs$drug) %>% sort()
# cefprozil maybe?
```