328: Variant selection boosts R2 for haptoglobin (HP) in cis‑Mendelian randomization

Zhou A et al., Human Genetics and Genomics Advances - Comparing LD‑pruning, COJO, SuSiE and PCA in haptoglobin (HP) cis‑region data, the study finds including non‑lead variants substantially increases variance explained (R2) and MR precision. Key terms: haptoglobin, cis-Mendelian randomization, LD-pruning, SuSiE, COJO. Study Highlights:The study analyzed circulating haptoglobin (HP) using Fenland protein GWAS summary statistics with LD from UK Biobank, compared four variant selection methods (modified LD‑pruning, COJO, SuSiE, PCA), and extended results with simulations and 15 additional gene regions. In the HP region, incorporating non‑lead variants produced a median proportional gain in R2 of 145.1% and a median reduction in MR standard error of 36.3% relative to the lead variant alone. In simulations with one or two causal variants the methods recovered the expected genetic variance (≈40%) and, when causal variants were removed, non‑lead‑inclusive methods recovered more variance than lead‑only. The functional implication supported by the data is that including correlated non‑lead variants can materially increase instrument strength and precision in cis‑MR, but may raise risks of pleiotropy and numerical instability. Conclusion:Variant selection methods that incorporate correlated non‑lead variants reliably improve instrument strength (R2) and MR precision in cis‑MR compared with the lead‑variant‑only approach; comparisons with the lead variant are advised to detect instability. Music:Enjoy the music based on this article at the end of the episode. Article title:Variant selection to maximize variance explained in cis-Mendelian randomization First author:Zhou A Journal:Human Genetics and Genomics Advances DOI:10.1016/j.xhgg.2026.100573 Reference:Zhou A, Karhunen V, Tian H, Pott J, Patel A, Slob EAW, Burgess S. Variant selection to maximize variance explained in cis-Mendelian randomization. Human Genetics and Genomics Advances. 2026 Apr 9;7:100573. https://doi.org/10.1016/j.xhgg.2026.100573. License:This episode is based on an open-access article published under the Creative Commons Attribution 4.0 International License (CC BY 4.0) - https://creativecommons.org/licenses/by/4.0/ Support:Base by Base – Stripe donations: https://donate.stripe.com/7sY4gz71B2sN3RWac5gEg00 Official website https://basebybase.com On PaperCast Base by Base you’ll discover the latest in genomics, functional genomics, structural genomics, and proteomics. Episode link: https://basebybase.com/episodes/hp-variant-selection-cis-mr QC:This episode was checked against the original article PDF and publication metadata for the episode release published on 2026-03-27. QC Scope:- article metadata and core scientific claims from the narration- excludes analogies, intro/outro, and music- transcript coverage: Audited the transcript sections describing: (1) the four variant selection methods and their rationale; (2) HP region results including R2 gains and SE reductions; (3) simulation studies with known causal variance (40%); (4) extension to 15 gene regions; (5) pleiotropy concerns and safeguards; (6) practical recommendat- transcript topics: Four variant selection methods (LD-pruning, COJO, SuSiE, PCA); Modified LD-pruning with adjusted R2 and LD-matrix checks; HP region results: variance explained (R2) gains and MR precision; Simulations with known causal variance (40%); Two-causal-variant scenario and lead variant variance explained; Extension to 15 additional gene regions QC Summary:- factual score: 10/10- metadata score: 10/10- supported core claims: 8- claims flagged for review: 0- metadata checks passed: 4- metadata issues found: 0 Metadata Audited:- article_doi- article_title- articl... Chapters (00:00:20) - How a single matrix can cripple genomics(00:01:34) - Deep Dive: The Search for genetic instruments without breaking the math(00:05:59) - The Hidden Problem with Standard LD Pruning(00:11:13) - The Lead Variants vs Non-Lead variants in disease prediction(00:15:27) - Multivariant Analysis: The Right Mix of Variants