Query OXFORDIA

This page is for researchers who want to run analyses across the OXFORDIA federation from R.

Prerequisites

• R with the oxfordia and solidauthr packages installed
• A WebID issued by your institution's OXFORDIA identity provider
• The dataset URLs for the partner institutions you want to query (provided by their data administrators)
• A Statistic Access Rule at each partner site that authorizes your WebID for the statistics you want to run

Installation

install.packages("oxfordia")
install.packages("solidauthr")

Or install from the project repository:

remotes::install_github("OXFORDIA-project/oxfordia-r")
remotes::install_github("OXFORDIA-project/solidauthr-r")

---

Step 1: Authenticate

Authenticate once. Your WebID credential is presented automatically to every node you query.

library(oxfordia)
library(solidauthr)

auth <- solid_login(idp = "https://oxfordia.med.ox.ac.uk")

Replace the idp URL with your own institution's identity provider.

---

Step 2: Declare target sites

List the datasets you want to query. Each name is an arbitrary label; each value is the dataset URL provided by the partner site's data administrator.

targets <- oxfordia_targets(
  oxford    = "https://oxfordia.med.ox.ac.uk/cohort/nemaline",
  hacettepe = "https://oxfordia.hacettepe.edu.tr/cohort/nemaline",
  partner3  = "https://oxfordia.partner3.edu/cohort/nemaline"
)

---

Step 3: Run a query

The query is dispatched to all target sites in parallel. Each site independently checks your identity and access rules, executes the query locally, and returns an aggregate result. The R client combines per-site results into a single global answer.

Mean

result <- oxfordia_mean(
  targets    = targets,
  auth       = auth,
  graph_path = "BaselineAge"
)

result$value    # 21.6309888889  — global weighted mean
result$n        # 180            — total participant count
result$per_site # tibble with columns: site, mean, count

Kaplan–Meier survival analysis

result <- oxfordia_kaplan_meier(
  targets       = targets,
  auth          = auth,
  time_path     = "EventTime",
  event_path    = "EventOccurred",
  stratify_path = "GeneticGroup"   # optional
)

plot(result)   # federated survival curve, grouped by GeneticGroup

The plot() method produces a survival curve aggregated across all responding sites, with optional confidence intervals and stratification.

---

What happens behind the scenes

1. The R client builds an aggregate query and dispatches it to each target node.
2. Each node verifies your identity using your WebID token.
3. Each node loads the Statistic Access Rules applicable to you and the requested graph path, and checks whether the query is permitted.
4. If authorized, the node executes the query against its local triplestore.
5. After execution, the node checks result-level constraints (e.g., minCount). If the constraint fails, that site returns no result.
6. Each site returns its local aggregate result and a count (the weight used for combining).
7. The R client combines per-site results — for means, a count-weighted average; for Kaplan–Meier, a pooled curve.

If a site rejects your query (because no matching SAR exists, or minCount was not met), that site is omitted from the combined result. result$per_site shows you which sites contributed.

---

Handling partial results

Not every site in your targets list will necessarily respond or authorize every query. Check result$per_site to see per-site breakdown and identify any sites that did not contribute.

result$per_site
# A tibble: 3 × 3
#   site      mean     n
#   <chr>    <dbl> <int>
# 1 oxford    22.1    60
# 2 hacettepe 20.8    40
# 3 partner3  21.9    80

---

Available graph paths

Graph paths are defined by the Data Plugin installed on each node. For the Nemaline plugin, the available paths are:

Path	Description
BaselineAge	Age at enrollment
EventTime	Time to event or censoring
EventOccurred	Event indicator
GeneticGroup	Genetic subgroup
Ambulation	Ambulatory status
TotalMFM	Total MFM score

If a node does not have the required Data Plugin installed, or if your WebID is not authorized for the requested path, that site will return a rejection.