Evidence-structure analysis for human iPSC-neuron research: reconciling PubMed literature, SpliceAI, and population genetics to prevent single-lab false positives.
Operates at the unit of the individual experiment record. Identifies independent lab counts, assay diversity, and leave-one-lab-out stability so single-lab findings never outrank replicated results.
Directly cross-checks the literature against SpliceAI predictions, gnomAD constraint, ClinVar, and DepMap, surfacing published phenotypes that contradict population genetics.
Ships with 114 automated regression tests encoding biological constraints, proactive API-contract monitoring against Europe PMC / PubMed drift, and zero silent data coercion.
# 1. Install SYNAPTIK
git clone https://github.com/Fr4n0b/synaptik.git && cd synaptik
pip install -e .
# 2. Explore synthetic knowledge base
synaptik demo --db data/demo.db
# 3. Analyze phenotype associations with leave-one-out robustness
synaptik phenotype "SYN1 puncta density" --db data/demo.db -v
# 4. Launch local Flask web interface
synaptik ui --db data/demo.db
# Serving at http://127.0.0.1:8420from synaptik.store import Store
from synaptik.analyst import Analyst
from synaptik.schema import Direction
analyst = Analyst(Store("data/synaptik.db"))
result = analyst.analyze_phenotype("SYN1 puncta density", direction=Direction.DECREASE)
for assoc in result.strong_only():
print(assoc.pathway_label, f"{assoc.n_labs} independent labs", assoc.confidence_tier())