In inrae/Rodam: Wrapper Functions for 'ODAM' (Open Data for Access and Mining) Web Services

knitr::opts_chunk$set(echo = TRUE, warning=FALSE, fig.align=TRUE)

Demonstration of the functionnalities of the R ODAM package

Description

• 'ODAM' (Open Data for Access and Mining) is an experimental data table management software to make research data accessible and available for reuse with minimal effort on the part of the data provider. Designed to manage experimental data tables in an easy way for users, ODAM provides a model for structuring both data and metadata that facilitates data handling and analysis. It also encourages data dissemination according to FAIR principles by making the data interoperable and reusable by both humans and machines, allowing the dataset to be explored and then extracted in whole or in part as needed.

• The Rodam package has only one class, odamws that provides methods to allow you to retrieve online data using 'ODAM' Web Services. This obviously requires that data are implemented according the 'ODAM' approach , namely that the data subsets were deposited in the suitable data repository in the form of TSV files associated with their metadata also described in TSV files.

• The R ODAM package offers a set of functions for retrieve data and their metadata of datasets that are implemented help with the "Experimental Data Table Management System" (EDTMS) called ODAM, which stands for "Open Data for Access and Mining".

• See https://inrae.github.io/ODAM/ for further information.

Load the R ODAM package

library(Rodam)

Initialize the ODAM object

Initialize the 'ODAM' object with the wanted dataset along with its corresponding URL of webservice

dh <- new('odamws', wsURL='https://pmb-bordeaux.fr/getdata/', dsname='frim1')

Get the Data Tree

options(width=256)
options(warn=-1)
options(stringsAsFactors=FALSE)

show(dh)

Get all WebService entries

Get all WebService entries defined in the data subset 'samples'

dh$getWSEntryByName("samples")

NOTE:

a 'WSEntry' is an alias name associated with an attribute that allows user to query the data subset by putting a filter condition (i.e. a selection constraint) on the corresponding attribute. Not all attributes have a WSEntry but only few ones, especially the attributes within the identifier and factor categories. For instance, the WSEntry of the 'SampleID' attribute is 'sample'. Thus, if you want to select only samples with their ID equal to 365, you have to specify the filter condition as 'sample/365'.

Get data from 'samples' subset with a constraint

data <- dh$getDataByName('samples','sample/365')
data

But if this WSEntry concept is not clear for you, you can retrieve the full data subset, then performe a local selection as shown below :

data <- dh$getDataByName('samples') 
data[data$SampleID==365, ]

Convert all numeric values of date and time in a human-readable format

data$HarvestDate <- dh$dateToStr(data$HarvestDate)
data$HarvestHour <- dh$timeToStr(data$HarvestHour)
data[data$SampleID==365, ]

Get 'activome' data subset

Get 'activome' data subset along with its metadata

ds <- dh$getSubsetByName('activome')
ds$samples   # Show the identifier defined in the data subset
ds$facnames  # Show all factors defined in the data subset

options(width=150)

ds$varnames  # Show all quantitative variables defined in the data subset
ds$qualnames # Show all qualitative variables defined in the data subset
ds$WSEntry   # Show all WS entries defined in the data subset

Boxplot of all variables defined in ds$varnames

Rank <- simplify2array(lapply(ds$varnames, function(x) { round(mean(log10(ds$data[ , x]), na.rm=T)) }))
cols <- c('red', 'orange', 'darkgreen', 'blue', 'purple')
boxplot(log10(ds$data[, ds$varnames]), outline=F, horizontal=T, border=cols[Rank], las=2, cex.axis=0.8)

Find how many IDs in common there are between the subsets

Based on the subset network, the common ID to be considered is the "SampleID" identifier

 refID <- "SampleID"
 subsetList <- c( "samples", "activome", "qNMR_metabo", "cellwall_metabo" )
 n <- length(subsetList)
 Mintersubsets <- matrix(data=0, nrow=n, ncol=n)
 for (i in 1:(n-1))
     for (j in (i+1):n)
          Mintersubsets[i,j] <- length(dh$getCommonID(refID,subsetList[i],subsetList[j]))

 rownames(Mintersubsets) <- subsetList
 colnames(Mintersubsets) <- subsetList
 Mintersubsets[ -n, -1 ]

Get the merged data of two data subsets based on their common identifiers

setNameList <- c("activome", "qNMR_metabo" )
dsMerged <- dh$getSubsetByName(setNameList)

Boxplot of all variables defined in ds$varnames

cols <- c( rep('red', length(dsMerged$varsBySubset[[setNameList[1]]])), 
           rep('darkgreen', length(dsMerged$varsBySubset[[setNameList[2]]])) )
boxplot(log10(dsMerged$data[, dsMerged$varnames]), outline=F, horizontal=T, border=cols, las=2, cex.axis=0.8)

---

R Session Information

options(width=128)
sessionInfo()

inrae/Rodam documentation built on July 27, 2022, 1:36 a.m.