README

Description

Small shiny app to perform visual validation of CNV calls. It features four different modes:

1. Normal: the app will go through all CNVs in the provided table
2. Simple Filtering: the CNVs table will be filtered based on the provided criteria (vo, GT, length, numsnp).
3. Fixed locus: the app will select putative carriers in the provided locus. If any filter is provided, it will also be applied (before selecting). NB: in this mode the locus is the units of analysis, not the CNVs.
4. Select CNVs in region: the app will allow the user to specify a genomic region and will only show CNVs that overlap with this region (above a certain IOU threshold). Also in this mode any provided filter will be applied first.

Inputs

This app takes as input:

• a folder to save results
• a table of CNVs (sample_ID, chr, start, end, GT, vo, CN, length, numsnp)
• a table linking each sample_ID to a tabix-indexed file (sample_ID, file_path_tabix)
• a table of filtered SNP (Name, Chr, Position)

The tabix-indexed files should contain is expected to be a tab-separated file with the following columns (no header): chr, position, position, LRR, BAF, LRR_adj (if LRR_adj is not present, the app will use LRR instead). The input files are in the usual format for my packages and pipelines, described in details here.
Example for all input files are provided in data.

Citation

If you use this tool in your work please cite "Accurate and Effective Detection of Recurrent Copy Number Variants in Large SNP Genotype Datasets", DOI: https://doi.org/10.1002/cpz1.621.

Interface

The app interface is divided in two main parts:

• A sidebar on the left with the controls and input for filtering, fixed locus, and select region. The sidebar can be hidden by clicking on the small arrow at the top left.
• The main panel on the right shows the CNV line as a table on the top and the plot in the middle of the screen. Below the plot there are buttons to navigate through CNVs, store the visual validation for the CNV and to refine boundaries. The plot has two rows. BAF is on the top, LRR on the bottom. Each dot represent a SNP. Semi-transparent rectangles indicate the CNV call(s) in the region, while a gray box represents the locus of region in fixed locus or select region mode. In all modes the app will plot all CNVs for the sample (before any filtering) on the same chromosome as the current CNV. The CNV of interest is highlighted with a thin border. In fixed locus and select region modes, the locus/region is also highlighted with a gray box and thicker border.

Below the plot there are buttons to navigate through CNVs, store the visual validation for the CNV and to refine boundaries. The buttons are:

• True: confirm the current CNV call
• False: reject the current CNV call
• Unknown: mark the current CNV call as unclear
• Error: mark the current CNV call as erroneous
• Previous: go to the previous CNV
• Next: go to the next CNV
• Update CNV coordinates: update current CNV coordinates, see next section for details.

The buttons can also be activated using the keyboard:

• T: True
• F: False
• U: Unknown
• E: Error
• P: Previous
• N: Next
• R: Refine CNV coordinates

Interact with the plot

This new version uses plotly to display the CNVs. This allows to dinamically change the view of the plot, both zooming in and out, and panning left and right. The zoom can be controlled with a dedicated slider in the sidebar. By default, the click and drag is set to "select" mode, to be used for the boundaries function. To move (pannig) the plot, select the pan icon in the top right corner of the plot (4 arrows pointing outwards).

Update Boundaries

This new version implements also the ability to refine CNV boundaries. First the user needs to select a region in the plot (both LRR or BAF work). The text below the validation buttons will display the new coordinates and the updated number of markers. Note that this is is based on the displayed dots so it will change depending on whetther all markers are shown or just the filtered ones. If the user is satisfied with the selection, they can click on the Update CNV coordinates button. This will update the current CNV info and mark it as True. Any other CNV overlapping the new boundaries will be marked as '-1' and will show a warning message when visualised.

How to launch the app

To use the app just run Rscript app.R /path/to/workingdir /path/to/cnvs.txt /path/to/samplee.txt /path/to/snps.txt. Input files can be anywhere, all path must be full (starting from / with no symlinks), the results will be saved in workingdir. Every times the app the app moves from one CNV to the next (or previous) the output table is saved in the provided workingdir as project_name.tsv. If workingdir the app will attempt to create it. If the also the parent folder does exist, the app will assume there is a typo in the provided path and will stop.
NB: the app will overwrite any existing project_name.tsv file in the provided workingdir without warning.

Visual inspection codes

The vo column is coded as follows:

• 1 : true
• 2 : false
• 3 : unknown/unclear
• -9 : new
• -7 : error

How to install the app

Just clone the repo, app.R is all is needed to run the program (assuming the dependencies are met).

R dependencies:

• shiny
• data.table
• ggplot2
• bslib
• DT
• plotly

Linux dependencies:

• tabix
• a browser

OLD versions

An old version of the app (and of this README) is available in the old_version folder. This version is simpler and uses a static plot, for this reason it should also be faster. However, it does not allow to zoom in and out of the plot, and it does not have the boundary refinement feature. It is also less polished in general.