2020.05.11_report.Rmd

---
title: "Project update"
author: "Kim Dill-McFarland, kadm@uw.edu"
date: "version `r format(Sys.time(), '%B %d, %Y')`"
output: pdf_document
subtitle: J. Shah chimeric mouse collaboration
editor_options:
  chunk_output_type: console
---

## Completed

1. Change gene-level model to contrasts of WT:TKO within `uninfected` or `infected` samples
2. Compare above contrasts model to previous interaction term model (`~status*cell`) 
    - Roughly half of genes identified as significant in either model are also significant in the other model.
    - Genes unique to the interaction model are only significant for infection status. Genes unique to the contrasts model were only significant for cell type within `infected` cells. 
    - Thus, the contrasts model appears to be successfully removing much of the infection-only signal while amplifying cell type signal.
3. Modules made from 4950 genes significant (FDR < 0.3) in contrasts model
    - Resulted in 18 modules + 399 genes in module 0
4. Contrasts model of modules
    - All modules significant at FDR < 0.05
    - 3 modules significant in uninfected, 5 in infected (ignore module 0), and 10 for both
5. GSEA of significant genes and genes in each module

## To discuss

* Other methods for assigning function to modules
* Additional analyses?

## To-do

* GSEA of fold change module groups (*e.g.* up in uninfected, up in infected, up in both, etc)
* Heatmap of module expression
* Modify boxplots for publication
* Point group to count files

***