UMAP3Dviz.Rmd

---
title: "3D UMAP reduction and visualization"
author: "Clemente Calabrese"
date: "2023-04-14"
output: html_document
---

```{r setup, message = FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

# Data and Tools loading
## Load the libraries needed
```{r message = FALSE}
library(tidyverse)
library(Seurat)
library(SeuratObject)
library(patchwork)
```
# Clusterization
## Load the experimental integrated dataset and run the clusterization scripts
These functions generate PCA and UMAP coordinates (used for dimensionality reduction); also they split cells into clusters, based on their similarity.
```{r, message = FALSE}
df <- RunPCA(df, verbose = FALSE)
df <- RunUMAP(df, reduction = "pca", dims = 1:30, n.components = 3L) #n.neighbors=30, min.dist = 0
df <- FindNeighbors(df, reduction = "pca", dims = 1:30)
df <- FindClusters(df, resolution = c(0.4,1))  #CLUSTER RESOLUTIONS(0.4,1)

```

## Retrieve the 3D UMAP coordinates
Save the coordinates of each cell to a new file named "UMAP_coord.tsv"
```{r}
as.data.frame(df[["umap"]]@cell.embeddings) %>%
  write_delim(file = "UMAP_coord.tsv", delim = "\t")
```
Later this file will be retrieved by the Blender API to plot it in the 3D space