---
title: "GNOSIS: Genomics explorer using statistical and survival analysis in R"
author:
- name: Lydia King
  affiliation: [
    "University of Galway, Ireland"
  ]
package: GNOSIS
output:
  BiocStyle::html_document:
    toc: false
vignette: >
  %\VignetteIndexEntry{GNOSIS Overview}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

# Introduction

Exploratory, statistical and survival analysis of cancer genomic data is 
extremely important and can lead to new discoveries, such as the identification 
of novel genomic prognostic markers, that have the potential to advance our 
understanding of cancer and ultimately benefit patients. These analyses are 
often performed on data available from a number of consortia websites, such as 
cBio Cancer Genomics Portal (cBioPortal), which is one of the best known and
commonly used consolidated curations that hosts data from large consortium 
efforts. While cBioPortal provides both graphical user interface (GUI)-based and 
representational state transfer mediated means for researchers to explore and 
analyse clinical and genomics data, its capabilities have their limitations and 
oftentimes, to explore specific hypotheses, users need to perform a more 
sophisticated ‘off site’ analysis that typically requires users to have some 
prior programming experience.  

To overcome these limitations and provide a GUI that facilitates the 
visualisation and interrogation of cancer genomics data, particularly 
cBioPortal-hosted data, using standard biostatistical methodologies, we 
developed an R Shiny app called GeNomics explOrer using StatistIcal and Survival 
analysis in R (GNOSIS). GNOSIS was initially developed as part of our study, 
using the METABRIC data, to investigate whether survival outcomes are associated 
with genomic instability in luminal breast cancers and was further developed to 
enable the exploration, analysis and incorporation of a diverse range of genomic 
features with clinical data in a research or clinical setting. 

GNOSIS leverages a number of R packages and provides an intuitive GUI with 
multiple tab panels supporting a range of functionalities, including data upload 
and initial exploration, data recoding and subsetting, data visualisations, 
statistical analysis, mutation analysis and, in particular, survival analysis 
to identify prognostic markers. In addition, GNOSIS also helps researchers carry 
out reproducible research by providing downloadable input logs (Shiny_Log.txt)
from each session. 

GNOSIS has been submitted to Bioconductor to aid researchers in carrying out a 
reproducable, comprehensive statistical and survival analysis using data 
obtained from cBioPortal, or otherwise.

# Installation

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

BiocManager::install("GNOSIS")
```

# Loading the package

```{r,include=TRUE, results="hide", message=FALSE, warning=FALSE}
library(GNOSIS)
```

# Launching GNOSIS

```{r, eval=FALSE}
GNOSIS()
```

# GNOSIS layout 

The GNOSIS GUI has 4 main elements:
(1) A sidebar where each analysis tab can be selected, the Exploratory Tables tab
    is selected and displayed.
(2) Tab panels within each tab, allowing multiple operations to be carried out
    and viewed in the one tab.
(3) A box sidebar allowing users to select inputs, alter arguments and customise
    and export visualisations.
(4) Main viewing panel displaying output.

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Pic.png")
```

# Data upload and preview 

Users can upload their own clinical, CNA or mutation data stored on their local
machine, or select a cBioPortal study to upload:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_File_Upload.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Select_Study.png")
```

<br>

A preview of the uploaded/selected data is provided in the GNOSIS viewing panel 
to ensure that the data has been read in correctly:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Preview_Clinical.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Preview_CNA.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Preview_MAF.png")
```
<br>

In the case where a cBioPortal study does not contain CNA and/or MAF data, a 
warning will be produced alerting users to this. 

In addition, users can select specific columns of each dataframe to inspect:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Column_Selection.png")
```

# Data reformatting and filtering

To prepare the data for downstream analysis a number a things can be done. 
Firstly users can change the type of variables to numeric or factors using the 
box sidebar, which contains a space to select relevant variables:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Reformatting.png")
```

<br>

Subsequently, users can subset the data based on up to three categorical 
variables and carry out survival variable recoding.

Here we filter the data to only include patients who received chemotherapy:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Filtering.png")
```

We also recode the overall and disease-specific survival to 0/1: 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Survival_Recoding.png")
```

<br>
 
In cases where CNA data is uploaded, users may produce and segment CNA metrics 
for each patient, as well as select and extract specific genes for further 
analysis: 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_CNA_Calc.png")
```

# Data visualisation 

Users can produce a range of visualisations including boxplots, scatterplots, 
barplots, histograms and density plots.

Here is an example of a customisable boxplot, that can also be downloaded:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Boxplot.png")
```

# Statistical and survival analysis 

The primary function offered by GNOSIS is statistically robust survival analysis. 
GNOSIS contains several step-wise tabs to provide a complete survival analysis 
of the data under investigation.

Users can produce KM survival curves and the corresponding logrank tests to 
identify survival-associated categorical variables, both visually and 
statistically. 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_KM_Curves_Clinical.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_KM_Curves_CNA.png")
```

<br>

Users can perform a selection of association tests to identify variables that 
are associated with each other. This enables users to identify potential 
confounding variables in the analysis. 

Statistical association tests available include the Chi-squared test, Fisher’s 
exact test, simulated Fisher’s exact test, ANOVA, Kruskal-Wallis test, 
pairwise t-test and Dunn’s test. 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Association.png")
```

<br>

Users can produce both univariate and multivariable Cox models to identify 
survival-associated variables, and test the assumptions of these models using 
graphical diagnostics based on the scaled Schoenfeld residuals: 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Univariate_Cox.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Multivariable_Cox.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Cox_Assumption.png")
```

<br>

The corresponding adjusted survival curves, survival curves adjusted for the 
covariates in the multivariable Cox model, can also be produced and customised:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Adjusted_KM_Curves.png")
```

<br>

In the case where the PH assumption of the multivariable Cox model is violated, 
users can apply recursive partitioning survival trees: 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Rpart_Tree.png")
```

# Mutation Analysis

An additional function of GNOSIS is the ability to summarise, analyse and 
visualise mutation annotation format (MAF) files using maftools. 

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_MAF_Summary_Text.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_MAF_Summary_Plot.png")
```
<br>
```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_MAF_Lollipop_Plot.png")
```

# Input Log

GNOSIS facilitates reproducible research by allowing users to download an input 
log containing information on all the inputs selected throughout the session:

```{r,echo=FALSE}
knitr::include_graphics("figures/GNOSIS_Input_Log.png")
```

# Additional Resources

For details on the implementation, layout and application of GNOSIS see the 
corresponding [publication](https://hrbopenresearch.org/articles/5-8#). 
Demonstration videos providing a walkthrough of GNOSIS are also provided on 
[Zenodo](https://zenodo.org/record/5788659). 

# Session Info

```{r}
sessionInfo()
```