Identify druggable synthetic lethal paralogs

PhD project code for systematically identifying approved small-molecule drugs associated with synthetic lethal paralogs

Overview

druggable_paralogs is the code accompanying my PhD work on systematic identification of small-molecule drugs associated with synthetic lethal paralog targets. It integrates approved-drug and target data from Open Targets, bioactivity measurements from ChEMBL, and paralog synthetic lethality predictions to pinpoint paralog pairs that are both druggable and therapeutically relevant.

What it does

  • Preprocesses paralog pair data from De Kegel et al. (2021), integrating sequence and structural similarity scores and mapping identifiers across HGNC, Ensembl, and UniProt.
  • Characterises the landscape of approved small-molecule drugs that target paralog pairs, and tests for paralog enrichment within each protein family using Fisher’s exact test.
  • Retrieves pChEMBL and pKi bioactivity measurements from ChEMBL 36, classifies approved drugs by their paralog-targeting pattern (single-member vs. co-targeting), and compares potency across paralog pairs.
  • Integrates druggable pairs with synthetic lethal predictions and combinatorial CRISPR screen data to prioritise therapeutically relevant paralog pairs.

Why it matters

Paralog synthetic lethality is a promising route to expand the druggable cancer genome, but translating predicted vulnerabilities into therapies depends on whether tractable small-molecule chemistry already exists. This project bridges that gap by cross-referencing predicted SL paralog pairs with the approved-drug and bioactivity landscape, surfacing pairs where a therapeutic starting point is already in hand.