12-confounders-reads.Rmd

---
knit: bookdown::preview_chapter
---

# Identifying confounding factors (Reads)

```{r, echo=FALSE, message=FALSE, warning=FALSE}
library(scater, quietly = TRUE)
library(knitr)
options(stringsAsFactors = FALSE)
opts_chunk$set(out.width='90%', fig.align = 'center', echo=FALSE)
reads <- readRDS("tung/reads.rds")
reads.qc <- reads[fData(reads)$use, pData(reads)$use]
endog_genes <- !fData(reads.qc)$is_feature_control
```

```{r confound-pca-reads, fig.cap = "PCA plot of the tung data"}
scater::plotPCA(reads.qc[endog_genes, ],
                colour_by = "batch",
                size_by = "total_features",
                exprs_values = "log2_counts")
```

```{r confound-find-pcs-total-features-reads, fig.cap = "PC correlation with the number of detected genes", fig.asp=1}
scater::plotQC(reads.qc[endog_genes, ],
               type = "find-pcs",
               variable = "total_features",
               exprs_values = "log2_counts")
```

```{r confound-find-expl-vars-reads, fig.cap = "Explanatory variables"}
scater::plotQC(reads.qc[endog_genes, ],
               type = "expl",
               exprs_values = "log2_counts",
               variables = c("total_features",
                             "total_counts",
                             "batch",
                             "individual",
                             "pct_counts_feature_controls_ERCC",
                             "pct_counts_feature_controls_MT"))
```