---
title: "User's Guide"
author: "Shian Su"
output:
    BiocStyle::html_document:
        toc: true
vignette: >
  %\VignetteIndexEntry{User's Guide}
  %\VignetteEngine{knitr::rmarkdown}
  \usepackage[utf8]{inputenc}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
knitr::opts_chunk$set(fig.width = 7)
knitr::opts_chunk$set(fig.height = 5)
knitr::opts_chunk$set(dpi = 72)

# preload to avoid loading messages
library(NanoMethViz)
library(dplyr)
exon_tibble <- get_exons_mm10()
```

# Introduction

The `NanoMethViz` package is designed to visualise methylation data from long-read sequencing. It provides a simple interface to plot methylation data, and to perform dimensionality reduction on the data. The package is designed to work with methylation data on a per-read basis.

We provide functions for converting to various formats for downstream analysis, and visualising methylation data at various levels of resolution. Features are all developed to work with 5mC CpG methylation data, but it will work with other modifications if the data is in the correct format, as the type of modification is not generally specified in the data.

This package is designed to work with methylation data from the Oxford Nanopore Technologies (ONT) platform. Currently it supports data from the following software:

- dorado (modBAM)
- modkit (tsv generated by `modkit extract full`)
- Megalodon (tsv)
- Nanopolish (tsv)
- f5c (tsv)

It is possible to use data from other software as long as it matches the format of the data used in this package. If you would like any further other formats supported please create an issue at https://github.com/Shians/NanoMethViz/issues.

# Installation

To install this package, run

```{r, eval = FALSE}
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("NanoMethViz")
```

```{r}
library(NanoMethViz)
library(dplyr)
```

# Importing data

In order to use this package, your data must be converted from the output of methylation calling software to a tabix-indexed bgzipped format. The data needs to be sorted by genomic position to respect the requirements of the samtools [tabix](http://www.htslib.org/doc/tabix.html) indexing tool. On Linux and macOS systems this is done using the bash `sort` utility, which is memory efficient, but on Windows this is done by loading the entire table and sorting within R.

## modBAM data

In the latest versions of `dorado` and other ONT software, the modifications are stored in the BAM file. This can be used directly with `NanoMethViz` using the `ModBamResult` class as a substitute for `NanoMethResult`.

This requires a `ModBamFiles` object, which is a list of paths to the BAM files and the sample names. The sample names must match the sample names in the `sample_anno` data frame.

Because the BAM files contain significantly more data than the tabix files, it may be useful to convert the BAM files to tabix files using the  `modbam_to_tabix()` function for sharing the data.

In modBAM files the modification calls are made along the read, allowing sequencing errors to produce modifications calls where there is none in the genome. This may lead to noise impacting the averaging results. The `NanoMethViz.site_filter` option can be used to remove sites with low coverage, by default it is set to 3 to remove any sites with coverage less than 3. See the "Site filtering" section for more information.

```{r}
# construct with a ModBamFiles object as the methylation data
mbr <- ModBamResult(
    methy = ModBamFiles(
        samples = "sample1",
        system.file(package = "NanoMethViz", "peg3.bam")
    ),
    samples = data.frame(
        sample = "sample1",
        group = 1
    ),
    exons = exon_tibble
)

# use in the same way as you would a NanoMethResult object
plot_gene(mbr, "Peg3", heatmap = TRUE)
```


## Tabular data (Modkit, Megalodon, Nanopolish, f5c)

Tabular data is the most common format for methylation data, and is the format used by most of the software. The columns in the tabular data varies according to the software used. `NanoMethViz` will attempt to automatically detect the format of the data.

### Modkit

Modkit can be used to extract site-level modification data modBAM files. The `modkit extract full` command will output a tab-separated file with the required columns that can be converted useing the `create_tabix_file()` function.

NOTE: `modkit` BED files are not supported as they do not contain read-level information, only position-aggregated data.

## Megalodon

To import data from Megalodon's modification calls, the [per-read modified bases](https://nanoporetech.github.io/megalodon/file_formats.html#per-read-modified-bases) file must be generated. This can be done by either adding `--write-mods-text` argument to Megalodon run or using the `megalodon_extras per_read_text modified_bases` utility.

Later version of Megalodon will also produce `modBAM` files, which can be used directly with `NanoMethViz` using the `ModBamResult` class as a substitute for `NanoMethResult`.

## Importing to tabix format

Once you have produce the correct output from the software listed above, the conversion can be done using the `create_tabix_file()` function. We provide example data of nanopolish output within the package.

```{r}
methy_calls <- system.file(package = "NanoMethViz",
    c("sample1_nanopolish.tsv.gz", "sample2_nanopolish.tsv.gz"))
```

We then create a temporary path to store a converted file, this will be deleted once you exit your R session. Once `create_tabix_file()` is run, it will create a .bgz file along with its tabix index. Because we have a small amount of data, we can read in a small portion of it for inspection, do not do this with large datasets as it decompresses all the data and will take very long to run.

## Importing data

```{r, message=F}
# create a temporary file to store the converted data
methy_tabix <- file.path(tempdir(), "methy_data.bgz")
samples <- c("sample1", "sample2")

# you should see messages when running this yourself
create_tabix_file(methy_calls, methy_tabix, samples)

# don't do this with full datasets, it will take a long time
# we have to use gzfile to tell R that we have a gzip compressed file
methy_data <- read.table(
    gzfile(methy_tabix), col.names = methy_col_names(), nrows = 6)

methy_data
```

Now `methy_tabix` will be the path to a tabix object that is ready for use with NanoMethViz.

# Quick start

## Constructing the NanoMethResult object

To generate a methylation plot we need 3 components:

* methylation data in tabix format
* annotation of exons
* annotation of samples

The methylation information has been modified from the output of nanopolish/f5c. It has then been compressed and indexed using `bgzip()` and `indexTabix()` from the `Rsamtools` package. Here `system_file()` is used to retrieve data internal to the package, for your down data you simply need to provide a path without the `system_file()` function.


```{r}
# methylation data stored in tabix file
methy <- system.file(package = "NanoMethViz", "methy_subset.tsv.bgz")

# tabix is just a special gzipped tab-separated-values file
read.table(gzfile(methy), col.names = methy_col_names(), nrows = 6)
```

The exon annotation was obtained from the `Mus.musculus` package, and joined into a single table. It is important that the chromosomes share the same convention as that found in the methylation data.

```{r}
# helper function extracts exons from TxDb package
exon_tibble <- get_exons_mm10()

head(exon_tibble)
```

We will defined the sample annotation ourselves. It is important that the sample names match those found in the methylation data.

```{r}
sample <- c(
  "B6Cast_Prom_1_bl6", "B6Cast_Prom_1_cast",
  "B6Cast_Prom_2_bl6", "B6Cast_Prom_2_cast",
  "B6Cast_Prom_3_bl6", "B6Cast_Prom_3_cast"
)

group <- c("bl6", "cast", "bl6", "cast", "bl6", "cast")

sample_anno <- data.frame(sample, group, stringsAsFactors = FALSE)

sample_anno
```

For convenience we assemble these three pieces of data into a single object.

```{r}
nmr <- NanoMethResult(methy, sample_anno, exon_tibble)
```

The genes we have available are

* Peg3
* Meg3
* Impact
* Xist
* Brca1
* Brca2

## Basic Plotting

For demonstrative purposes we will plot Peg3. The plot on top shows the average methylation probabilities across all samples for each experimental condition. The heatmap below it shows the methylation probabilities along individual reads, with disjoint reads stacked onto the same row. The annotation down the bottom shows the isoforms of the Peg3 gene as well as its directionality.

```{r}
plot_gene(nmr, "Peg3")
```

We can also load in some DMR results to highlight DMR regions.

```{r}
# loading saved results from previous bsseq analysis
bsseq_dmr <- read.table(
    system.file(package = "NanoMethViz", "dmr_subset.tsv.gz"),
    sep = "\t",
    header = TRUE,
    stringsAsFactors = FALSE
)
```

```{r}
plot_gene(nmr, "Peg3", anno_regions = bsseq_dmr)
```

See other plotting options in the [package documentation](https://shians.github.io/NanoMethViz/reference/index.html).

# Exporting data

The methylation data can be exported into formats appropriate for bsseq, DSS, or edgeR.

## bsseq and DSS

Both bsseq and DSS make use of the BSSeq object, and these can be obtained from the NanoMethResult objects using the `methy_to_bsseq()` function.

```{r, message = FALSE}
nmr <- load_example_nanomethresult()
bss <- methy_to_bsseq(nmr)

bss
```

## edgeR

edgeR can also be used to perform differential methylation analysis: https://f1000research.com/articles/6-2055. BSSeq objects can be converted into the appropriate format using the `bsseq_to_edger()` function. This can be used to count reads on a per-site basis or over regions.

```{r}
gene_regions <- exons_to_genes(NanoMethViz::exons(nmr))
edger_mat <- bsseq_to_edger(bss, gene_regions)

edger_mat
```

# Importing Annotations

This package comes with helper functions that import exon annotations from the Bioconductor packages `Homo.sapiens` and `Mus.musculus`. The functions `get_exons_hg19()` and `get_exons_mm10()` simply take data from the respective packages, and reorganise the columns into the following columns:

-   gene_id
-   chr
-   strand
-   start
-   end
-   transcript_id
-   symbol

This is used to provide gene annotations for the gene or region plots.

For other annotations, they will most likely be able to be imported using `rtracklayer::import()` and manipulated into the desired format. As an example, we can use a small sample of the C. Elegans gene annotation provided by ENSEMBL. `rtracklayer` will import the annotation as a `GRanges` object, this can be coerced into a data.frame and manipulated using `dplyr`.

```{r}
anno <- rtracklayer::import(system.file(package = "NanoMethViz", "c_elegans.gtf.gz"))

head(anno)
```

```{r}
anno <- anno %>%
    as.data.frame() %>%
    dplyr::rename(
        chr = "seqnames",
        symbol = "gene_name"
    ) %>%
    dplyr::select("gene_id", "chr", "strand", "start", "end", "transcript_id", "symbol")

head(anno)
```

## Importing Annotations

Annotations can be simplified if full exon and isoform information is not required. For example, gene-body annotation can be represented as single exon genes. For example we can take the example dataset and transform the isoform annotations of Peg3 into a single gene-body block. The helper function `exons_to_genes()` can help with this common conversion.

```{r, message = FALSE}
nmr <- load_example_nanomethresult()

plot_gene(nmr, "Peg3")
```

```{r}
new_exons <- NanoMethViz::exons(nmr) %>%
    exons_to_genes() %>%
    mutate(transcript_id = gene_id)

NanoMethViz::exons(nmr) <- new_exons

plot_gene(nmr, "Peg3")
```

# Dimensionality reduction

Dimensionality reduction is used to represent high dimensional data in a more tractable form. It is commonly used in RNA-seq analysis, where each sample is characterised by tens of thousands of gene expression values, to visualise samples in a 2D plane with distances between points representing similarity and dissimilarity. For RNA-seq the data used is generally gene counts, for methylation there are generally two relevant count matrices, the count of methylated bases, and the count of methylated bases. The information from these two matrices can be combined by taking log-methylation ratios as done in Chen et al. 2018.

### Preparing data for dimensionality reduction

It is assumed that users of this package have imported the data into the gzipped tabix format. From there, further processing is required to create the log-methylation-ratio matrix used in dimensionality reduction. Namely we go through the BSseq format as it is easily coerced into the desired matrix and is itself useful for various other analyses.

```{r, message = FALSE}
# convert to bsseq
bss <- methy_to_bsseq(nmr)
bss
```

We can generate the log-methylation-ratio based on individual methylation sites or computed over genes, or other feature types. Aggregating over features will generally provide more stable and robust results, here we will use genes.

```{r}
# create gene annotation from exon annotation
gene_anno <- exons_to_genes(NanoMethViz::exons(nmr))

# create log-methylation-ratio matrix
lmr <- bsseq_to_log_methy_ratio(bss, regions = gene_anno)
```

NanoMethViz currently provides two options, a MDS plot based on the limma implementation of MDS, and a PCA plot using BiocSingular.

```{r}
plot_mds(lmr) +
    ggtitle("MDS Plot")

plot_pca(lmr) +
    ggtitle("PCA Plot")
```

Additional coloring and labeling options can be provided via arguments to either function. Further customisations can be done using typical ggplot2 commands.

```{r}
new_labels <- gsub("B6Cast_Prom_", "", colnames(lmr))
new_labels <- gsub("(\\d)_(.*)", "\\2 \\1", new_labels)
groups <- gsub(" \\d", "", new_labels)

plot_mds(lmr, labels = new_labels, groups = groups) +
    ggtitle("MDS Plot") +
    scale_colour_brewer(palette = "Set1")
```

Points can also be plotted without labels by setting `labels = NULL`.

```{r}
plot_mds(lmr, labels = NULL, groups = groups) +
    ggtitle("MDS Plot") +
    scale_colour_brewer(palette = "Set1")
```

# Package options

## Site filtering

NanoMethViz offers a site filtering option to remove sites with low coverage. This is particularly useful for modBAM files as the modifications calls are made along the reads, allowing methylation to be called at miscalled sites. The site filter is set to 3 by default, to filter any sites with coverage less than 3. This can be changed using the option `NanoMethViz.site_filter`. The following will remove any sites with coverage less than 5 from queries and plots.

```{r, eval = FALSE}
options("NanoMethViz.site_filter" = 5)
```

## Region annotation colours

By default the `anno_regions` argument in plotting functions will create bands coloured grey, specifically `grey50`. This can be changed globally across the package using the option `NanoMethViz.highlight_col`. For example the following will change the colour to red.

```{r, eval = FALSE}
options("NanoMethViz.highlight_col" = "red")
```