README.md
In lmuenter/oracl:

Lukas Muenter 27 7 2021

oracl

NOTE: Currently, this package does only accept AGI-codes (A. thaliana). This will change, however.

This package provides a client for GO-Term enrichment via the API of PANTHER. It takes a vector of gene IDs, sends it to PANTHER, and reformats the response into a handy dataframe. This dataframe also includes gene IDs, which are associated to the GO-Term in question.

# install from github
devtools::install_github("lmuenter/oracl")

In this example, we’d like to identify overrepresented GO-Terms for an example dataset provided with the package. Note, that we specify the Biological Process ontology by setting ont = bp in oracl::oraclient(). Other options are of course ont = mf (Molecular Function) and ont = cc (Cellular Component).

# load package
library(oracl)

# Get a set of AGI-codes.
gs <- oracl:::GS01

# Get a background geneset (optional)
bg <- oracl:::background

# conduct GO-Term ORA via PANTHER
bp.df <- oraclient(gs, bg = bg, ont = "bp", fdr.thresh = 0.05)

## Joining, by = "GO_ID"

# Load Packages
library(ggplot2)

# Make a plot
volcano.p = volcanoracl(bp.df)

## Adding missing grouping variables: `grouping`

## Joining, by = c("label", "grouping")

# The plot `volcano.p` is a ggplot-object.
# We can change its attributes!
volcano.p + scale_colour_gradientn(colours = "steelblue")

Using `{oracl}` with a list of genesets

When several genesets should be inferred, it may be handy to combine overrepresented terms in one dataframe. This is especially useful for plotting.

# obtain a list of genesets
gs.ls <- list(
    oracl:::GS01,
    oracl:::GS02,
    oracl:::GS03
)

# get background geneset
bg <- oracl:::background

# set names of list elements (vital for later)
names(gs.ls) <- c("GS01", "GS02", "GS03")

# get overrepresented GO-terms
bp.ls = lapply(gs.ls, oraclient, 
    bg = bg,
    ont = "bp",
    fdr.thresh = 0.05
)

## Joining, by = "GO_ID"
## Joining, by = "GO_ID"
## Joining, by = "GO_ID"

# get ONE dataframe (ID-column `grouping` specifies the geneset)
bp.ls.df <- oracl_list_to_df(bp.ls)

We can now plot overrepresented GO-Terms using group information in the column bp.df$grouping. Here, we want to facet the plot according to the grouping variable (stored in bp.df$grouping). We also specify the desired number of columns, the position of the facet label, and whether or not we only include labels found in each dataset (change these things according to your data!):

oraclot(bp.ls.df, top_n = 5) + 
    facet_wrap(grouping ~ ., ncol = 1, strip.position = "right", scales = "free_y") +
    scale_color_viridis_c()

## Adding missing grouping variables: `grouping`

## Joining, by = c("label", "grouping")

We can also make a facetted volcano plot:

volcanoracl(bp.ls.df, top_n = 5) +
    facet_wrap(grouping ~ ., nrow = 1)

## Adding missing grouping variables: `grouping`

## Joining, by = c("label", "grouping")

Gene IDs and Organism. Currently, only Arabidopsis thaliana (L.) Heynh. can be investigated.
Cognate genes. in order to save resources, the API of PANTHER does not report gene sets back (personal communication). Gene IDs reported by {oracl} are therefore only approximations. In essence, the underlying geneset is semantically compared to a gene-to-GO-term-dataset for every enriched GO-Term. These datasets are included in {oracl} (see oracl/data/goterms). Datasets have been generated by conducting ORA using the PANTHER website with all available AGI codes. To obtain necessary datasets, all results (without Bonferroni Correction) were exported to .json, parsed, and reformated.