sypres.io

📐 Website for the SYPRES Project

Website Overview

The site is organized into four main sections, accessible from the top navigation bar:

Section	URL	Description
Studies	/docs/datasets	Overview of SYPRES and living evidence, with individual study pages (e.g. Psilocybin for Depression)
Team	/docs/team	Current SYPRES team members
Publications	/docs/community/publications	Published papers and citations
News	/docs/blog	Project updates, conference appearances, and announcements

Additional pages linked from within the Studies section include search terms, extraction SOPs, effect sizes, and reproducibility guides for each dataset.

Analysis & Project Setup

For setting up the R environments used in SYPRES analyses, see the analysis README.

Scripts

Python scripts in scripts/ handle reference formatting and enrichment. They require Python 3.10+ and a few packages:

pip install pybtex requests

format_pubs.py — Project publication formatting

Converts SYPRES's own project publication BibTeX files (under _data/publications/*.bib) into HTML citation lists (under _includes/publications/). These are the publications by the SYPRES team, not the trial reference lists.

python scripts/format_pubs.py

Reads from and writes to:

• _data/publications/core_publications.bib → _includes/publications/core_publications.html
• _data/publications/psilodep.bib → _includes/publications/psilodep.html
• _data/publications/mdmaptsd.bib → _includes/publications/mdmaptsd.html

parse_references.py — Trial reference lists (with Blossom enrichment)

Generates numbered HTML reference lists for each living review dataset. If Blossom API enrichment data exists (see below), it merges per-study summaries and abstracts as expandable <details> blocks under each citation.

python scripts/parse_references.py

Reads from and writes to:

• docs/_data/psilodep_refs.json + _data/publications/psilodep_blossom.json → _includes/publications/psilodep_refs.html
• docs/_data/mdmaptsd_refs.json + _data/publications/mdmaptsd_blossom.json → _includes/publications/mdmaptsd_refs.html

If no *_blossom.json file exists for a dataset, the script still works — it just renders plain citations without expanders.

fetch_blossom_metadata.py — Blossom API enrichment

Fetches per-study metadata (paragraph summaries, abstracts, study characteristics) from the Blossom API and writes enrichment JSON files that parse_references.py consumes.

Setup: Export your Blossom API key as an environment variable (never commit it):

export BLOSSOM_API_KEY=ak_your_key_here

Usage:

# Fetch all datasets:
python scripts/fetch_blossom_metadata.py --all

# Fetch a single dataset:
python scripts/fetch_blossom_metadata.py --dataset psilodep

# Dry-run a single study (prints raw API response, writes nothing):
python scripts/fetch_blossom_metadata.py --dataset psilodep --only "Davis, 2021" --dry-run

# Force re-fetch (ignore 30-day cache):
python scripts/fetch_blossom_metadata.py --dataset psilodep --force

The script caches responses locally in scripts/.blossom_cache/ (gitignored) with a 30-day TTL. Re-runs are fast and idempotent.

Full regeneration workflow

When references or enrichment data change, regenerate everything with:

export BLOSSOM_API_KEY=ak_...
python scripts/fetch_blossom_metadata.py --all
python scripts/parse_references.py
python scripts/format_pubs.py

Development

Building and serving locally

If you haven't installed ruby:

brew install rbenv ruby-build
rbenv install 3.1.4 # current version of ruby the website uses
eval "$(rbenv init -)"

You only need to run this command once:

bundle install

Then you can serve locally anytime using:

bundle exec jekyll serve

If your terminal session is restarted, you may need to re-run the eval command when you come back. Then you can run bundle exec jekyll serve again and it should work. If you're having trouble, check that the version is correct: ruby -v.