Creating custom alternative splicing annotation

psichomics quantifies alternative splicing based on alternative splicing event annotations from MISO, SUPPA, VAST-TOOLS and rMATS. New alternative splicing annotation may be prepared and used in psichomics by parsing alternative splicing events from those tools. Please contact me if you would like to see support for other tools.

This tutorial will guide you on how to parse alternative splicing events from different tools. To do so, start by loading the following packages:

library(psichomics)
library(plyr)

SUPPA annotation

SUPPA generates alternative splicing events based on a transcript annotation. Start by running SUPPA’s generateEvents script with a transcript file (GTF format) for all event types, if desired. See SUPPA’s page for more information.

The resulting output will include a directory containing tab-delimited files with alternative splicing events (one file for each event type). Hand the path of this directory to the function parseSuppaAnnotation(), prepare the annotation using prepareAnnotationFromEvents() and save the output to a RDS file:

# suppaOutput <- "path/to/SUPPA/output"

# Replace `genome` for the string with the identifier before the first
# underscore in the filenames of that directory (for instance, if one of your 
# filenames of interest is "hg19_A3.ioe", the string would be "hg19")
suppa <- parseSuppaAnnotation(suppaOutput, genome="hg19")
annot <- prepareAnnotationFromEvents(suppa)

# suppaFile <- "suppa_hg19_annotation.RDS"
saveRDS(annot, file=suppaFile)

rMATS annotation

Just like SUPPA, rMATS also allows to generate alternative splicing events based on a transcript annotation, although two BAM or FASTQ files are required to generate alternative splicing events. Read rMATS’ page for more information.

The resulting output of rMATS is then handed out to the function parseMatsAnnotation():

# matsOutput <- "path/to/rMATS/output"
mats <- parseMatsAnnotation(
    matsOutput,         # Output directory from rMATS
    genome = "fromGTF", # Identifier of the filenames
    novelEvents=TRUE)   # Parse novel events?
annot <- prepareAnnotationFromEvents(mats)

# matsFile <- "mats_hg19_annotation.RDS"
saveRDS(annot, file=matsFile)

MISO annotation

Simply retrieve MISO’s alternative splicing annotation and give the path to the downloaded folder as input.

# misoAnnotation <- "path/to/MISO/annotation"
miso <- parseMisoAnnotation(misoAnnotation)
annot <- prepareAnnotationFromEvents(miso)

# misoFile <- "miso_AS_annotation_hg19.RDS"
saveRDS(annot, file=misoFile)

VAST-TOOLS annotation

Download and extract VAST-TOOLS’ alternative splicing annotation and use the path to the TEMPLATES subfolder as the input of parseVastToolsAnnotation(). Complex events (i.e. alternative coordinates for the exon ends) are not currently supported.

# vastAnnotation <- "path/to/VASTDB/libs/TEMPLATES"
vast <- parseVastToolsAnnotation(vastAnnotation, genome="Hsa")
annot <- prepareAnnotationFromEvents(vast)

# vastFile <- "vast_AS_annotation_hg19.RDS"
saveRDS(annot, file=vastFile)

Combining annotation from different sources

To combine the annotation from different sources, provide the parsed annotations of interest simultaneously to the function prepareAnnotationFromEvents:

# Combine the annotation from SUPPA, MISO, rMATS and VAST-TOOLS
annot <- prepareAnnotationFromEvents(suppa, vast, mats, miso)

# annotFile <- "AS_annotation_hg19.RDS"
saveRDS(annot, file=annotFile)

Quantifying alternative splicing using the custom annotation

The created alternative splicing annotation can be used in psichomics for alternative splicing quantification. To do so, when using the GUI version of psichomics, be sure to select the Load annotation from file… option, click the button that appears below and select the recently created RDS file.

Otherwise, if you are using the CLI version, perform the following steps:

annot <- readRDS(annotFile) # "annotFile" is the path to the annotation file
junctionQuant <- readFile("ex_junctionQuant.RDS") # example set

psi <- quantifySplicing(annot, junctionQuant)

## Using 1 of 60 events (2%) whose junctions are present in junction quantification data...

psi # may have 0 rows because of the small junction quantification set

##                                                   Normal 1  Normal 2 Normal 3
## SE_1_+_23385660_23385840_23385851_23395032_KDM1A 0.7777778 0.4444444      0.6
##                                                  Cancer 1  Cancer 2  Cancer 3
## SE_1_+_23385660_23385840_23385851_23395032_KDM1A      0.4 0.4285714 0.5555556

Preparing an Alternative Splicing Annotation for psichomics

Nuno Saraiva Agostinho

2022-01-27