4_FASscore_analysis.Rmd

---
title: "FAS score analysis"
author:
- name: Mario Keller
  affiliation: Faculty of Biological Sciences, Goethe University Frankfurt
output:
  BiocStyle::html_document:
      toc: TRUE
      toc_float: TRUE
      code_folding: hide
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message=FALSE, warning=FALSE,
                      crop=NULL, results = TRUE)
```

```{r}
library(tidyverse)
library(knitr)
library(ggbeeswarm)
library(ggsci)
library(Biostrings)
library(plotly)
library(PhyloProfile)
```

```{r}
myTheme <- theme_bw() +
    theme(axis.text = element_text(size = 14, colour="black"),
          axis.title = element_text(size=16, colour="black"),
          axis.ticks=element_line(color="black"),
          axis.ticks.length=unit(.15, "cm"),
          panel.border=element_rect(color="black", fill = NA),
          panel.background = element_blank(),
          plot.background = element_blank(),
          legend.text = element_text(size=12),
          legend.position = "none")
```

# Background 

Changes in domain composition had been analyzed using the FAS algorithm, which
results in FAS scores between pairs of protein sequences. The results are 
analyzed in this report.

# Data

```{r input}
FASscores <- read.table("data/FAS/regulated_isoforms_adjusted_IDs_no_redundant_IDs.phyloprofile",
                        header=T)

sequenceInformation <- readxl::read_xlsx("xlsx_files/summary_A5SS_adjusted_Swissprot_sequences.xlsx")

classificationDf <- readRDS("rds_files/classificationDF.rds")

seqs <- readAAStringSet("fasta_files/regulated_isoforms_adjusted_IDs_no_redundant_IDs.fasta")

domainFileFWD <- "data/FAS/regulated_isoforms_adjusted_IDs_no_redundant_IDs_forward.domains"
domainFileREV <- "data/FAS/regulated_isoforms_adjusted_IDs_no_redundant_IDs_reverse.domains"


domainDfFWD <- parseDomainInput(seed = NULL, domainFileFWD, "file")
domainDfREV <- parseDomainInput(seed = NULL, domainFileREV, "file")
```

The original FAS output looks like this:

```{r}
FASscores %>%
    dplyr::slice(1:10) %>%
    kable(., "html", align="c", caption="Fitted cassette exon events") %>%
    kableExtra::kable_styling("striped") %>%
    kableExtra::scroll_box(height="300px", width = "94%")
```

The data.frame is adjusted by adding an ID (starting at 1), the event_id,
frameshift information, stop codon information and the group (shortened,
elongated, disrupted). There is also a group called "other", which is the
case when the distal and proximal proteins were identical (both share
the same stop codon -> intron in 3'UTR). All pairs belonging to "other"
have a score of 1. Actually this group is a subset of "disrupted" as
both Proximal and Distal share a stop codon.

FAS scores are directional, which is why there are two FAS score columns 
that are named FAS_F and FAS_B. The first one is the score for the
comparison of the sequence in column 1 against the sequence in column 3, 
meaning Distal against Proximal. The latter score is the other direction. 
FAS_F was renamed to Distal_in_Proximal and FAS_B to Proximal_in_Distal.

In addition, the column *ofInterest* is added, which is TRUE if the group 
is either shortened or elongated and one of the two scores is $\le$ 0.9,
which indicates a substantial change in the feature architecture.

```{r}

FASscores$ID <- 1:nrow(FASscores)

FASscores$event_id <- NA
FASscores$event_id[grepl("ENSG", FASscores$geneID)] <- paste0("ENSG", FASscores$geneID[grepl("ENSG", FASscores$geneID)] %>% strsplit(., "ENSG", fixed=T) %>% sapply(., "[[", 2) %>% strsplit(., "-D") %>% sapply(., "[[",1))
FASscores$event_id[grepl("ENSG", FASscores$orthoID)] <- paste0("ENSG", FASscores$orthoID[grepl("ENSG", FASscores$orthoID)] %>% strsplit(., "ENSG", fixed=T) %>% sapply(., "[[", 2) %>% strsplit(., "-P") %>% sapply(., "[[",1))



FASscores <- left_join(FASscores, classificationDf)

FASscores$group[is.na(FASscores$group)] <- "other"

FASscores <- FASscores %>% dplyr::rename(., Distal_in_Proximal = FAS_F,
                                         Proximal_in_Distal = FAS_B) %>%
    mutate(group = factor(group, levels = c("shortened", "elongated",
                                            "disrupted", "other"))) %>%
    arrange(desc(group))



FASscores <- FASscores %>%
    mutate(ofInterest = ifelse(group %in% c("shortened","elongated") & (Distal_in_Proximal <= 0.9 | Proximal_in_Distal <= 0.9), TRUE, FALSE))
```

The adjusted FAS output looks like this:

```{r}
FASscores %>%
    dplyr::slice(1:10) %>%
    kable(., "html", align="c") %>%
    kableExtra::kable_styling("striped") %>%
    kableExtra::scroll_box(height="300px", width = "94%")
```

# 2D FAS score plot

The two FAS scores of each pair are used for a scatterplot and the 
`r sum(FASscores$ofInterest)` pairs labeled as *ofInterest* highlighted with
a red circle.

```{r}

FASscores %>%
arrange(desc(group)) %>%
ggplot(., aes(x=Distal_in_Proximal, y=Proximal_in_Distal)) +
    geom_point(size=3, mapping=aes(col=group, alpha=ofInterest)) +
    geom_point(data = . %>% dplyr::filter(ofInterest),
               shape=1, size=5, col="red") +
    scale_color_manual(values=c("disruped" = "lightgrey",
                                "other" = "lightgrey",
                                "shortened" = "#42B540FF",
                                "elongated" = "#0099B4FF")) +
    scale_alpha_manual(values=c("TRUE" = 1,
                                "FALSE" = .25)) +
    myTheme + theme(legend.position = "bottom",
    axis.text.x = element_text(angle=45, hjust=1, vjust=1)) +
    guides(color = guide_legend(title.position = "top", title.hjust = 0.5),
           alpha = guide_legend(title.position = "top", title.hjust = 0.5))
```

# Beeswarm plot

```{r}
plot_df <- FASscores %>%
    select(ID, Distal_in_Proximal, Proximal_in_Distal, group) %>%
    mutate(group = forcats::fct_recode(group, disrupted="other")) %>%
    rowwise %>%
    mutate(min_score = min(Distal_in_Proximal, Proximal_in_Distal)) %>%
    mutate(color_point = ifelse(min_score <= 0.9, "Yes", "No")) %>%
    ungroup

ggplot(plot_df, aes(x=group, y=min_score, color = color_point)) +
    geom_quasirandom() +
    geom_boxplot(alpha=.5, outlier.size = -1, col="black") +
    coord_cartesian(ylim=c(0,1)) +
    scale_color_manual(values=c("Yes" = "red",
                                "No" = "darkgrey")) +
    labs(y="Minimal FAS score per event", x="") +
    myTheme +
    theme(aspect.ratio = 2/1,
          axis.text.x = element_text(angle=45, hjust=1, vjust=1))

ggplot(plot_df %>% filter(group != "disrupted"), aes(x=group, y=min_score, color = color_point)) +
    geom_quasirandom() +
    coord_cartesian(ylim=c(0,1)) +
    scale_color_manual(values=c("Yes" = "red",
                                "No" = "darkgrey")) +
    labs(y="Minimal FAS score per event", x="") +
    myTheme +
    theme(aspect.ratio = 2/1,
          axis.text.x = element_text(angle=45, hjust=1, vjust=1))
    

    
```


# Events of interest

A function to plot the pairwise alignments of the paired 
protein sequences was created. The scrollbar below the alignments allows to
approximate the position of the A5SS event in the Domain architecture 
plots. 

```{r}
printAlignment <- function(id){
    pair <- seqs[c(FASscores$geneID[FASscores$ID == id], 
                   FASscores$orthoID[FASscores$ID == id])]
    alg <- pairwiseAlignment(pair[1], pair[2],
                             type="global")
    seq <- c(alignedPattern(alg), alignedSubject(alg))
    print(seq %>% as.character)
}
```

## Shortened 

```{r}
ids_shortened <- FASscores %>%
    dplyr::filter(ofInterest & group=="shortened") %>%
    dplyr::pull(ID)

name_to_id_shortened <- FASscores %>%
    dplyr::filter(ID %in% ids_shortened) %>%
    dplyr::select(ID,event_id) %>%
    left_join(., sequenceInformation %>%
                  dplyr::select(event_id, gene_name))

```

```{r results='asis'}
for(id in ids_shortened){
    cat("\n")
    cat('\n## Gene name = ', name_to_id_shortened$gene_name[name_to_id_shortened$ID == id], '\n')
    
    print(
        FASscores[FASscores$ID == id,] %>%
        kable(., "html", row.names = F) %>%
        kableExtra::kable_styling("striped") %>%
        kableExtra::column_spec(column = c(1,3), width_min = "300px") %>%
        kableExtra::scroll_box(height="120px", width = "94%")
    )
    
    knitr::knit_child( text=c(
    '```{r}',
    'printAlignment(id)',
    '```',
    ''
      ) , envir=environment() , quiet=TRUE) %>% cat(., sep="\n")
    
    seedID <- FASscores[FASscores$ID == id,] %>% pull(geneID)
    orthoID <- FASscores[FASscores$ID == id,] %>% pull(orthoID)
    info <- c(seedID, orthoID)
    
    plotFWD <- tryCatch(createArchiPlot(info, domainDfFWD, 9, 9),
                    error=function(e){
                      ggplot()
                    })
    
    plotREV <- tryCatch(createArchiPlot(info, domainDfREV, 9, 9),
                error=function(e){
                  ggplot()
                })
    
    #plotFWD <- createArchiPlot(info, domainDfFWD, 9, 9)
    
    cat('\n### Forward domain file', '\n')
    print(ggpubr::ggarrange(plotFWD))
    cat("\n")
    
    cat('\n### Reverse domain file', '\n')
    print(ggpubr::ggarrange(plotREV))
    cat("\n")
}

```

## Elongated 

```{r}
ids_elongated <- FASscores %>%
    dplyr::filter(ofInterest & group=="elongated") %>%
    dplyr::pull(ID)

name_to_id_elongated <- FASscores %>%
    dplyr::filter(ID %in% ids_elongated) %>%
    dplyr::select(ID, event_id) %>%
    left_join(., sequenceInformation %>%
                  dplyr::select(event_id, gene_name))
```

```{r results='asis'}
for(id in ids_elongated){
    cat("\n")
    cat('\n## Gene name = ', name_to_id_elongated$gene_name[name_to_id_elongated$ID == id], '\n')
    
    print(
        FASscores[FASscores$ID == id,] %>%
        kable(., "html", row.names = F) %>%
        kableExtra::kable_styling("striped") %>%
        kableExtra::column_spec(column = c(1,3), width_min = "300px") %>%
        kableExtra::scroll_box(height="120px", width = "94%")
    )
    
    knitr::knit_child( text=c(
    '```{r}',
    'printAlignment(id)',
    '```',
    ''
      ) , envir=environment() , quiet=TRUE) %>% cat(., sep="\n")
    
    seedID <- FASscores[FASscores$ID == id,] %>% pull(geneID)
    orthoID <- FASscores[FASscores$ID == id,] %>% pull(orthoID)
    info <- c(seedID, orthoID)
    
    plotFWD <- tryCatch(createArchiPlot(info, domainDfFWD, 9, 9),
                    error=function(e){
                      ggplot()
                    })
    
    plotREV <- tryCatch(createArchiPlot(info, domainDfREV, 9, 9),
                error=function(e){
                  ggplot()
                })
    
    #plotFWD <- createArchiPlot(info, domainDfFWD, 9, 9)
    
    cat('\n### Forward domain file', '\n')
    print(ggpubr::ggarrange(plotFWD))
    cat("\n")
    
    cat('\n### Reverse domain file', '\n')
    print(ggpubr::ggarrange(plotREV))
    cat("\n")
}

```

# Session Information

```{r}
sessionInfo()
```