Model1_noControls.Rmd

---
title: "model1_noControls"
author: "Katrien Quintelier"
date: "2023-06-20"
output: html_document
editor_options: 
  chunk_output_type: inline
---

```{r setup}
knitr::opts_knit$set(root.dir = rprojroot::find_rstudio_root_file()) # change wd from path dir to project dir for all chunks
getwd()
```

# Load environment
```{r}
source("Environment.R")
```

Goal: Normalize files_P1_C1 and files_P2_C1 using the cellTypeMarkers

# Train model
```{r}
# Create aggregate fcs files that will be used as a proxi for a control file
set.seed(1)
agg_P1_C1 <- AggregateFlowFrames(fileNames = files_P1_C1,
                                 cTotal = length(files_P1_C1)*10000) 
agg_P2_C1 <- AggregateFlowFrames(fileNames = files_P2_C1,
                                 cTotal = length(files_P2_C1)*10000)

# Save aggregates
write.FCS(agg_P1_C1, "Data/Preprocessed/Panel1/Cohort1/agg_P1_C1.fcs")
write.FCS(agg_P2_C1, "Data/Preprocessed/Panel2/Cohort1/agg_P2_C1.fcs")


# Train model
model1_withoutControl <- CytoNorm.train(files = flowSet(agg_P1_C1, agg_P2_C1),
                                        labels = c("P1", "P2"),
                                        channels = cellTypeChannels,
                                        transformList = NULL,
                                        seed = 1,
                                        verbose = TRUE,
                                        plot = TRUE,
                                        FlowSOM.params = list(nCells = 1e+06, 
                                                              xdim = 7, ydim = 7, 
                                                              nClus = 3, 
                                                              scale = FALSE))
saveRDS(object = model1_withoutControl, file = "RDS/model1_withoutControl.rds")
```

# Test the CVs
```{r}
# Evaluate the CVs
CVs1 <- testCV(fsom = model1_withoutControl$fsom, cluster_values = 3:44)
pdf("Results/Model1_overviewCV.pdf")
PlotOverviewCV(fsom = model1_withoutControl$fsom, cv_res = CVs1,show_cv = 0.8, max_cv = 1.5)
dev.off()
```

# Apply CytoNorm model
```{r}
# Normalize files
CytoNorm.normalize(model = model1_withoutControl,
                   files = c(files_P1_C1, files_P2_C1),
                   labels = c(rep("P1", length(files_P1_C1)),
                              rep("P2", length(files_P2_C1))),
                   transformList = NULL,
                   verbose = TRUE,
                   prefix = "",
                   transformList.reverse = NULL, 
                   outputDir = "Data/Normalized_cellType")
```

# Density plots
```{r}
# List files
files_P1_C1_norm <- list.files(path = "Data/Normalized_cellType", 
                               pattern = "ID[1-4]_Panel1_TP..fcs", full.names = TRUE)
files_P2_C1_norm <- list.files(path = "Data/Normalized_cellType", 
                               pattern = "ID[1-4]_Panel2_TP..fcs", full.names = TRUE)

# Make plots
p <- plotDensities(input = list("P1_C1" = agg_P1_C1,
                                "P2_C1" = agg_P2_C1,
                                "P1_C1_norm" = files_P1_C1_norm,
                                "P2_C1_norm" = files_P2_C1_norm),
                   channels = cellTypeChannels,
                   colors = batch_colors, 
                   model = model1_withoutControl)

# Save to pdf
pdf("Results/Model1_densities.pdf", height = 4*length(cellTypeChannels), width = 3*(length(p)-1)/(2*length(cellTypeChannels)))
p_ <- ggarrange(ggarrange(plotlist = p[1:length(p)-1], ncol = (length(p)-1)/(2*length(cellTypeChannels)), nrow = 2*length(cellTypeChannels)),
                p$legend, nrow = 2, heights = c(10,1))
print(p_)
dev.off()
```

# Spline plots
```{r}
pdf("Results/Model1_splines.pdf", height = 9, width = 12)
plotSplines(model = model1_withoutControl, channels = cellTypeChannels, groupClusters = TRUE)
plotSplines(model = model1_withoutControl, channels = cellTypeChannels)
dev.off()
```