cysteine_grantProposalFigures.Rmd

---
title: "cysteine_grantProposalFigures"
output: html_document
---
# making figures to send to Jonas for the cysteine project grant proposal

# format df for the figures
```{r}
df <- maxHitScoreDF_forGrantFigures
df$perc_of_ecoli_score <- NA
df$`Category by Percentage of E. coli Hit Score` <- NA

# iterate by gene to get E. coli score for each gene
for (gene in unique(df$gene)) {
  ecoli_score <- subset(df$hit_score, df$strain == "Escherichia_coli" & df$ gene == gene)
  
  # iterate by row in df to calcualte percentile `Category by Percentage of E. coli Hit Score` of E. coli score and add to new column in df
  for (strain in unique(df$strain)) {
    df$perc_of_ecoli_score <- ifelse(df$strain == strain & df$gene == gene, 
                                     (df$hit_score / ecoli_score)*100, 
                                     df$perc_of_ecoli_score)
  }
  for (strain in unique(df$strain)) {
    df$`Category by Percentage of E. coli Hit Score` <- ifelse(df$strain == strain & df$gene == gene & df$perc_of_ecoli_score <50, 
                          "<50%",
                          ifelse(df$strain == strain & df$gene == gene & 
                                   df$perc_of_ecoli_score  >= 50 & df$perc_of_ecoli_score <= 75,
                          "50-75%",
                           ifelse(df$strain == strain & df$gene == gene & 
                                   df$perc_of_ecoli_score  > 75,
                          ">75%", 
                          df$`Category by Percentage of E. coli Hit Score`)))
  }
}

# lcd is NA bc I dont have a model organism hit score for it yet
# df$perc_of_ecoli_score[df$gene == "lcd"] <- NA
# df$`Category by Percentage of E. coli Hit Score`[df$gene == "lcd"] <- NA

# reorder the levels of the `Category by Percentage of E. coli Hit Score` variable
df$`Category by Percentage of E. coli Hit Score` <- factor(df$`Category by Percentage of E. coli Hit Score`, c("<50%", "50-75%", ">75%"))

df$gene <- factor(df$gene, c("cysK", "malY", "metC", "cysM", "tnaA", "lcd"))

# done!
```

# make the figures!
```{r}
plot <- ggplot(df, aes(x = gene, fill = `Category by Percentage of E. coli Hit Score`)) +
  geom_bar() +
  ggtitle("Number of Strains with Hits for Cysteine-Utilization Genes") +
  ylab("number of strains") +
  labs(fill="Category by Percentage\nof E. coli Hit Score") +
  theme(legend.title=element_text(size=8), legend.key.height=unit(1.5, "cm"))

plot

# ggsave(file = "NumberOfHitsPerGene.pdf", plot = plot, device = "pdf", width = 7, height = 5)
```