scRNA_pipeline_v1

scRNA Pipeline v1 notes

Needed packages:
Seurat, dplyr, magrittr, pheatmap, RColorBrewer, scales, umap (need to figure out with python)
* need to set reticulate directory before importing packages *

Needed variables:
mouse/human cell cycle genes
mouse or human data

Workflow:
* steps 1-3 need to be done outside of functions *
1) load file (find file address(es) w/ list.files())
  a) if have more than one file then put as list
2) create Seurat object(s) 
  a) list2env to make objects 
  b) add any metadata to objects
3) merge objects together w/ cell-ids added to front
  a) only needed when have more than 1 file 

4) QC
  a) PercentageFeatureSet for mitochondrial genes (mt for mouse and MT for human)
  b) VlnPlot (nCount/nUMI, nFeature/nGene, percent.mito) --> save
  c) FeatureScatter (nCount vs. nFeature, nCount vs. percent.mito) --> save
5) based on QC, filter cells if needed
  a) subset 
6) Processing
  a) NormalizeData
  b) FindVariableFeatures --> save
  c) ScaleData w/ CellCycleScoring
    i) based on 2000 varaible genes --> faster
  d) PCA
    i) decide how many pc's to use
  ds) if think will have batch effect
    i) use appropriate batch correction software
  e) FindNeighbors based on pc's
  f) FindClusters
  g) RunUMAP based on pc's
    i) visualize on DimPlot
7) Labeling
  a) auto labeling
    i) function needed to be made
  b) manual labeling
    i) FeaturePlot/Dotplot
8) DE
  a) FindMarker (b/w 2 groups) or FindAllMarkers (all groups/clusters)
  b) DoHeatmap or pheatmap --> use data slot
    maybe scale.data if doesn't look good
* If need more filtering, go back to step 4 or remove trash clusters * 
9) Subsetting
  depends on what want to subset
  if want to find subclusters --> start at step 6
  if have already wanted clusters --> continue
  DE/heatmap/other things