Clustering of faecal calprotectin and CRP profiles obtained from the Lothian IBD Registry (LIBDR), a population cohort of patients with IBD receiving care from the NHS Lothian health board. This code was used to conduct analyses described in Analysis for Large-scale clustering of longitudinal faecal calprotectin and C-reactive protein profiles in inflammatory bowel disease by Constantine et al. If you re-use any this code for your own analyses, we would be very grateful if you cited our paper.
The analysis requires access to the igmm/cvallejo-predicct datastore. As the data originated from unconsented patient data, we are currently unable to share these datasets with external entities.
This repository comes with its own R package, found in libdr,
which provides documented versions of custom functions used in the analysis.
Please open this directory in Rstudio as a project and run
devtools::install("./libdr") (or the Rstudio tooling) to install the
{libdr} package required for the analyses.
The code expects smb://cmvm.datastore.ed.ac.uk/igmm/ to be mounted to
/Volumes/igmm/. On MacOS, this can be achieved from Finder by pressing cmd+k
(or Go->Connect to Server from the task bar) and entering the above address
which will automatically map to /Volumes/igmm/.
The pipeline is currently set-up to use Quarto v1.4.551, the "next generation" replacement for Rmarkdown. The installer for this specific version can be downloaded via this link for macOS.
The scripts used to generate the reports can be found in source. As the interim processed data file are already available on datastore, the reports can be ran in any order. However, the following order is intended:
The reports in pre should be ran first as this will generate files in
igmm/cvallejo-predicct/libdr/processed/which are then used in selection to compare model performance and select the most appriopiate models. post contains all post hoc analyses which explore associations between baseline variables/advanced therapy prescrbing and cluster assignment. Faecal calprotectin and CRP clustering is also considered for subjects meeting the criteria for both biomarkers.
Any rendered reports will be found in the docs directory. scripts
contains any miscellaneous scripts not used in the reports or passed to Eddie
(such as scripts which have been used to generate poster figures)
