eval_dist.Rmd

---
title: "TCRemp distance properties"
author: "M.S."
date: "2024-11-27"
output:
  pdf_document: default
  html_document: default
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(reshape2)
library(stringr)
library(ggplot2)
```

```{r}
data <- read_tsv("res_TRB.txt.gz") |>
  rename(from = id)  |>
  mutate(from = as.character(from)) |>
  melt() |>
  filter(grepl("cdr3",variable))
data$to <- str_split_fixed(data$variable, "_", 2)[,1]
data$variable <- NULL
data <- data |>
  mutate(from = paste0("x", from),
         to = paste0("x", to)) |>
  group_by(from) |>
  mutate(value.scaled = (value - mean(value)) / sd(value)) |>
  ungroup()
glimpse(data)
```

```{r}
data1 <- data |> filter(as.character(to) > as.character(from))

ggplot(data1, aes(x = value)) +
  geom_density()

ggplot(data1, aes(x = scale(value))) +
  geom_density()

qqnorm(scale(data1$value))
qqline(scale(data1$value))
```

```{r}
data.m <- data |> 
  dcast(from ~ to)
rownames(data.m) <- data.m$from
data.m$from <- NULL
data.m <- as.matrix(data.m)
```

```{r}
dists <- dist(data.m) |> 
  as.matrix() |>
  melt() |>
  rename(from = Var1, to = Var2, dist_eucl = value) |>
  mutate(from = as.character(from), to = as.character(to))
```

```{r}
data.m[data.m > 100] <- 100
data.m[data.m < -100] <- -100
dists2 <- dist(data.m) |> 
  as.matrix() |>
  melt() |>
  rename(from = Var1, to = Var2, dist_eucl2 = value) |>
  mutate(from = as.character(from), to = as.character(to))
```

```{r}
data.comb <- data1 |>
  merge(dists) |>
  merge(dists2)
```

```{r}
data.comb |>
  ggplot(aes(x = value.scaled, 
             y = dist_eucl)) +
  geom_point(alpha = 0.1) +
  geom_density_2d() +
  geom_smooth(method = "lm") +
  xlab("True distance") +
  ylab("TCRemp dist") +
  theme_bw()

data.comb |>
  ggplot(aes(x = value.scaled, 
             y = dist_eucl2)) +
  geom_point(alpha = 0.1) +
  geom_density_2d() +
  geom_smooth(method = "lm") +
  xlab("True distance") +
  ylab("TCRemp dist") +
  theme_bw()
```

```{r}
cor.test(data.comb$value.scaled, data.comb$dist_eucl, method = "pearson")
cor.test(data.comb$value.scaled, data.comb$dist_eucl2, method = "pearson")
```

```{r}
cor.test(data.comb$value.scaled, data.comb$dist_eucl, method = "spearman")
cor.test(data.comb$value.scaled, data.comb$dist_eucl2, method = "spearman")
```

```{r}
#FIN