---
title: "scCertify Workflow"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{scCertify Workflow}
  %\VignetteEngine{knitr::rmarkdown}
  \usepackage[utf8]{inputenc}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(
    collapse = TRUE,
    comment = "#>"
)
```

Introduction

scCertify is an R package for explainable confidence scoring
of single-cell RNA-seq annotations.

The framework integrates:

1.Marker enrichment scoring

2.Neighborhood agreement

3.Entropy uncertainty

4.Doublet-aware confidence scoring

5.Confidence calibration

6.Explainable confidence attribution

Load Libraries

```{r load-libraries}
library(scCertify)

library(Seurat)

library(SingleR)

library(celldex)

library(UCell)

library(scDblFinder)

library(SingleCellExperiment)
```

Load Example Dataset

```{r load-data}
data("pbmc_small")

pbmc_small
```

Preprocessing

```{r pre-processing}
pbmc_small <- NormalizeData(
  pbmc_small
)

pbmc_small <- FindVariableFeatures(
  pbmc_small
)

pbmc_small <- ScaleData(
  pbmc_small
)

pbmc_small <- RunPCA(
  pbmc_small
)

pbmc_small <- RunUMAP(
  pbmc_small,
  dims = 1:10
)
```

Convert to SingleCellExperiment

```{r convert-to-sce}
sce <- as.SingleCellExperiment(
  pbmc_small
)
```

SingleR Annotation

```{r SingleR-Annotation}
ref <- HumanPrimaryCellAtlasData()

pred <- SingleR(
  test = sce,
  ref = ref,
  labels = ref$label.main
)

pbmc_small$predicted_label <-
  pred$labels
```

Doublet Detection

```{r Doublet-Detection}
sce <- scDblFinder(
  sce
)

pbmc_small$doublet_score <-
  colData(sce)$scDblFinder.score

pbmc_small$doublet_class <-
  colData(sce)$scDblFinder.class
```

Marker Database

```{r Marker-Database}
markers <- list(
  "B_cell" = c(
    "MS4A1",
    "CD79A"
  ),
  "T_cells" = c(
    "CD3D",
    "IL7R"
  ),
  "Monocyte" = c(
    "LYZ",
    "S100A8"
  ),
  "NK_cell" = c(
    "NKG7",
    "GNLY"
  ),
  "DC" = c(
    "FCER1A",
    "CST3"
  )
)
```

Entropy Calculation

```{r Calculate-Entropy}
pbmc_small$entropy_score <-
  entropy_score(
    pred$scores
  )

pbmc_small$entropy_norm <- (
  pbmc_small$entropy_score -

    min(pbmc_small$entropy_score)
) / (
  max(pbmc_small$entropy_score) -

    min(pbmc_small$entropy_score)
)
```

Run scCertify

```{r Run-package}
pbmc_small <- cell_certify(
  pbmc_small,
  markers
)
```

Confidence Classes

```{r Confidence-classes}
table(
  pbmc_small$confidence_class
)
```

Confidence Score Visualization

```{r CS-visualization}
FeaturePlot(
  pbmc_small,
  features = "confidence_score"
)
```

Confidence Class Visualization

```{r CC-Visualization}
DimPlot(
  pbmc_small,
  group.by = "confidence_class"
)
```

Explain Confidence Attribution

```{r ECA}
cell_id <- colnames(
  pbmc_small
)[1]

explain_confidence(
  pbmc_small,
  cell_id
)
```

Session Information

```{r Session-info}
sessionInfo()
```