README.Rmd

---
output: github_document
---

<!-- README.md is generated from README.Rmd. Please edit that file -->

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
```

# mmsig

<!-- badges: start -->
<!-- badges: end -->

The goal of mmsig is to provide a flexible and easily interpretable mutational signature analysis tool. mmsig was developed for hematological malignancies, but can be extended to any cancer with a well-known mutational signature landscape. 

mmsig is based on an expectation maximization algorithm for mutational signature fitting and applies cosine similarities for dynamic error suppression as well as bootstrapping-based confidence intervals and assessment of transcriptional strand bias.

Citation:
Rustad, E.H., Nadeu, F., Angelopoulos, N. et al. mmsig: a fitting approach to accurately identify somatic mutational signatures in hematological malignancies. Commun Biol 4, 424 (2021). https://doi.org/10.1038/s42003-021-01938-0

## Installation

You can install the development version from [GitHub](https://github.com/) with:

``` r
# install.packages("devtools")
devtools::install_github("evenrus/mmsig")
```

# Example

This is a basic example which shows mmsig usage

```{r}
library(mmsig)

data(mm_5_col)
data(signature_ref)

```

## setting up the mutational signature reference

```{r}
# remove canonical AID (SBS84) for genome-wide analysis
# remove the platinum signature (SBS35) because the patients are not platinum exposed

sig_ref <- signature_ref[c("sub", "tri", "SBS1", "SBS2", "SBS5", "SBS8", 
                           "SBS9", "SBS13", "SBS18", "SBS-MM1")] 

head(sig_ref)
```
## Setting up the mutation data

```{r}
# subset three samples to reduce run time

mm_5_col_subset <- mm_5_col[mm_5_col$sample %in% c("MEL_PD26412a", "MEL_PD26411c", "PD26414a"),]
head(mm_5_col_subset)
```
## Perform mutational signature analysis


```{r}
# Bootstrapping large datasets with many iterations can significantly increase runtime. 

set.seed(1)

sig_out <- mm_fit_signatures(muts.input=mm_5_col_subset, 
                             sig.input=sig_ref,
                             input.format = "vcf",
                             sample.sigt.profs = NULL, 
                             strandbias = TRUE,
                             bootstrap = TRUE,
                             iterations = 20, # 1000 iterations recommended for stable results
                             refcheck=TRUE,
                             cos_sim_threshold = 0.01,
                             force_include = c("SBS1", "SBS5"),
                             dbg=FALSE) 

```

## Plot signature estimates

```{r}
plot_signatures(sig_out$estimate, 
                samples = T, 
                sig_order = c("SBS1", "SBS2", "SBS13", "SBS5", "SBS8", "SBS9", 
                              "SBS18", "SBS-MM1", "SBS35"))

```

## Plot bootstraping confidence intervals

```{r}
bootSigsPlot(sig_out$bootstrap)
```

## Transcriptional strand bias for SBS-MM1

```{r}
head(sig_out$strand_bias_mm1)
```