---
title: "Getting started with SNPkit"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Getting started with SNPkit}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

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

```{r setup}
library(SNPkit)
library(snpStats)
library(methods)
```

# Overview

**SNPkit** provides a set of S4 tools for reading, organising, summarising,
filtering and exporting single nucleotide polymorphism (SNP) genotype data.
The central data structure is `SNPDataLong`, which bundles a genotype matrix
(`snpStats::SnpMatrix`), a marker map (`data.frame`), and metadata about the
data source.

This vignette walks through the typical steps:

1.  Building an `SNPDataLong` object from a toy genotype matrix.
2.  Inspecting the object with `summary()`.
3.  Applying quality-control filters with `qcSNPs()`.
4.  Exporting the cleaned data for use with external tools.

All file output uses `tempdir()` so the example does not write to the
user's home filespace.

# Building an `SNPDataLong` object

We simulate a tiny dataset with 10 individuals and 10 SNPs.

```{r build}
set.seed(123)

raw_mat <- matrix(
  as.raw(sample(1:3, 100, replace = TRUE)),
  nrow = 10, ncol = 10
)
rownames(raw_mat) <- paste0("ind", 1:10)
colnames(raw_mat) <- paste0("snp", 1:10)

geno <- new("SnpMatrix", raw_mat)

map <- data.frame(
  Name       = colnames(geno),
  Chromosome = rep(1, 10),
  Position   = seq_len(10),
  stringsAsFactors = FALSE
)

snp_data <- new(
  "SNPDataLong",
  geno      = geno,
  map       = map,
  path      = tempfile(),
  xref_path = "chip1"
)

snp_data
```

# Inspecting the object

The `summary()` method returns a `summary.SNPDataLong` object that can be
printed for a human-readable description or queried programmatically.

```{r summary}
s <- summary(snp_data)
s$n_individuals
s$n_snps
s$prop_missing
print(s)
```

# Quality control

`qcSNPs()` applies a flexible set of filters. The `action` argument controls
whether the function returns a report of removed SNPs (`"report"`), a filtered
`SNPDataLong` (`"filter"`), or both.

```{r qc}
filtered <- qcSNPs(
  snp_data,
  min_snp_cr   = 0.8,
  min_maf      = 0.05,
  snp_mono     = TRUE,
  no_position  = TRUE,
  action       = "filter"
)
filtered
```

# Exporting

`savePlink()` and `saveFImpute()` write files to a user-supplied directory.
For this vignette we use `tempdir()`.

```{r export}
out_dir <- file.path(tempdir(), "snpkit_demo")
dir.create(out_dir, showWarnings = FALSE)

savePlink(
  filtered,
  path       = out_dir,
  name       = "demo",
  run_plink  = FALSE,
  chunk_size = 5
)
list.files(out_dir, pattern = "demo")
```

# Where to go next

See `?qcSNPs`, `?savePlink`, `?saveFImpute`, and `?runAnticlusteringPCA` for
details on the individual functions. Functions that wrap external software
(FImpute, PLINK, ADMIXTURE) require the corresponding binary to be installed
on the system.