---
title: "11-byssal_thread_by_sample"
output: html_document
date: "2024-12-12"
---
### Load libraries
```{r}
library(tidyverse)
library(ggplot2)
library(readr)
```

### Load in gene count matrix and gene loc/gene ID key
```{r}
gene_count_matrix_clean <- read_csv("/home/shared/8TB_HDD_02/graceleuchtenberger/Github/byssus-exp-analysis/output/gene_count_matrix_clean.csv")

loc_go_blast <- read_delim("/home/shared/8TB_HDD_02/graceleuchtenberger/Github/byssus-exp-analysis/output/LOC_GO_list.txt",delim = "\t")

colnames(loc_go_blast)[colnames(loc_go_blast) == "gene"] ="loc_id"
```

```{r}
gene_count_matrix_clean <- gene_count_matrix_clean %>%
  mutate(loc_id = sub(".*\\|", "", ...1)) %>% # Extract part after the pipe
  select(loc_id, everything()) 


# View the updated data frame
print(df)
```

```{r}
annot_GCM <- left_join(gene_count_matrix_clean, loc_go_blast[,c("loc_id", "Protein.names", "uniprot_accession.y", "Gene.Names", "Organism")], by= "loc_id")

annot_GCM <- annot_GCM %>%
  select(loc_id, Protein.names, uniprot_accession.y, Gene.Names, Organism, everything())
```

```{r}
foot_GCM <- annot_GCM %>%
  filter(grepl("foot", Protein.names, ignore.case = TRUE))

byss_GCM <- annot_GCM %>%
  filter(grepl("byss", Protein.names, ignore.case = TRUE))

foot_GCM <- foot_GCM %>%
  distinct(uniprot_accession.y, .keep_all = TRUE)

byss_GCM <- byss_GCM %>%
  distinct(uniprot_accession.y, .keep_all = TRUE)

foot_byss_GCM <- rbind(foot_GCM, byss_GCM)


write.csv(foot_byss_GCM , file = "/home/shared/8TB_HDD_02/graceleuchtenberger/Github/byssus-exp-analysis/output/foot_byss_GCM.csv")
```

```{r}
foot_GCM_1 <- foot_byss_GCM %>%
  pivot_longer(!1:6, names_to = "sample", values_to = "count")

colnames(treatmentinfo.2)[colnames(treatmentinfo.2) == "sample.1"] ="sample"

foot_byss_GCM_2 <- left_join(foot_GCM_1, treatmentinfo.2, by = "sample")

write.csv(foot_byss_GCM_2, file = "/home/shared/8TB_HDD_02/graceleuchtenberger/Github/byssus-exp-analysis/output/foot_byss_GCM.csv")
```

```{r}
foot_byss_GCM_2 <- read_csv("/home/shared/8TB_HDD_02/graceleuchtenberger/Github/byssus-exp-analysis/output/foot_byss_GCM.csv")
```


```{r}
foot_byss_gene_plot <- foot_byss_GCM_2 %>%
  filter(tissue == "F") %>%
  ggplot(aes(x=Protein.names, y=log(count), fill=treatment))+
  geom_boxplot()+
  theme_classic()+
  theme(axis.text.x = element_text(angle = 45, hjust = 1))+
  theme(plot.margin = unit(c(0.5, 0.5, 0.5, 1), 
                                "inches"))
foot_byss_gene_plot
  
ggsave(foot_byss_gene_plot, file = "output/foot_byss_gene_plot.pdf")

```


```{r}
gill_byss_gene_plot<- foot_byss_GCM_2 %>%
  filter(tissue == "G") %>%
  ggplot(aes(x=Protein.names, y=log(count), fill=treatment))+
  geom_boxplot()+
  theme_classic()+
  theme(axis.text.x = element_text(angle = 45, hjust = 1))+
  theme(plot.margin = unit(c(0.5, 0.5, 0.5, 1), 
                                "inches"))+
  ylim(1,10)

gill_byss_gene_plot


ggsave(gill_byss_gene_plot, file = "output/gill_byss_gene_plot.pdf")
```