tsne_and_pca_look.Rmd

---
title: "T-Sne and PCA Look"
author: "D. Ford Hannum Jr."
date: "7/30/2020"
output: 
        html_document:
                toc: true
                toc_depth: 3
                number_sections: false
                theme: united
                highlight: tango
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, message = FALSE)
library(Seurat)
library(ggplot2)
library(data.table)
library(MAST)
```

```{r loading data}
wbm <- readRDS('./data/wbm_clustered_filtered_named.rds')
#DimPlot(wbm, reduction = 'umap', label = T, repel = T) + NoLegend()
```

```{r changing the levels of the data}
new_levels <- c('Granulocyte','Granulocyte','Granulocyte', 'B-cell','MEP', 
        'Granulocyte', 'Granulocyte','Monocyte','Macrophage','Erythroid','B-cell',
        'T-cell/NK', 'MK')
names(new_levels)  <- levels(wbm)
#new_levels
wbm <- RenameIdents(wbm, new_levels)
```

```{r tsne}
wbm <- RunTSNE(wbm, dims = 1:10)
```

```{r 3a pallette}

color_pal <- c("#0072B2", "#CC79A7", "#009E73", "#56B4E9","#D55E00",
               "#E69F00","#999999", "black")

DimPlot(wbm, reduction = 'tsne', 
        cols = color_pal, ncol = 2, pt.size = .01) +
        theme_bw() +
        theme(text = element_text(size = 10, family = 'sans'))
```

Doesn't look any better than the UMAP plot, so sticking with the UMAP plot makes the most sense.


```{r pca dimension reduction}
DimPlot(wbm, reduction = 'pca', 
        cols = color_pal, ncol = 2, pt.size = .01) +
        theme_bw() +
        theme(text = element_text(size = 10, family = 'sans'))

DimPlot(wbm, reduction = 'pca', split.by = 'condition', 
        cols = color_pal, ncol = 2, pt.size = .01) +
        theme_bw() +
        theme(text = element_text(size = 10, family = 'sans'))

DimPlot(wbm, reduction = 'pca', split.by = 'state', 
        cols = color_pal, ncol = 2, pt.size = .01) +
        theme_bw() +
        theme(text = element_text(size = 10, family = 'sans'))
```

I find it interesting that PC2 seems to be separating the MEPs from all other cells.


```{r looking at PC2}
print(wbm[['pca']], dims = 2, nfeatures = 10)

VizDimLoadings(wbm, dims = 2, reduction = 'pca')

DimHeatmap(wbm, dims = 2, cells = 500, balanced = T)
```