LoginSign Up
← Back to all sparks
R

RMVMR

ANALYTICS
Velocity0.0

RMVMR

github.com/WSpiller/RMVMR ↗

RMVMR is being tidied in lockstep with MVMR, the package it wraps

mendelian randomizationrradial methodsgeneticsrefactoringcausal inference
◆Current state
RMVMR provides radial multivariable Mendelian randomization — radial IVW estimation and plots layered over the MVMR package's conditional instrument-strength machinery. It has no independent release schedule: versions arrive alongside MVMR's, pin a minimum MVMR version, and fix defects in the seam between the two. Its most recent release landed the same day as new versions of MVMR and OneSampleMR.
◆Where it's heading
The work is code hygiene with results held fixed. ivw_rmvmr() now fits the radial IVW model explicitly rather than inheriting variables left over from the orientation loop, an undocumented data element carrying unused intermediate frames is gone, and plot_rmvmr() stops recomputing univariate radial analyses it already has — roughly halving the RadialMR calls with identical output. Each note is explicit that coefficients, standard errors and degrees of freedom are unchanged, which is a deliberate contrast with MVMR's own 2026 releases, where several fixes did change reported values.
◆Prediction
Because the package pins MVMR versions rather than vendoring behaviour, the next release most likely follows MVMR's next correctness fix; nothing in the notes points to independent feature work.

◆Recent moves

  1. 1mo ago

    RMVMR fixes a gencov error and halves redundant radial calls

    Four changes, all careful to leave estimates alone. strength_rmvmr() no longer errors when gencov is supplied as a matrix or vector instead of the scalar default; plot_rmvmr() stops recomputing the univariate radial analyses under the default cordat = NULL, roughly halving RadialMR::ivw_radial() calls with unchanged results; ivw_rmvmr() fits the canonical radial IVW model directly rather than through leftover orientation-loop variables; and an undocumented data element holding unused intermediates is dropped. Published the same day as MVMR's own correctness release.

    View source ↗
  2. 3mo ago

    RMVMR 0.4.4

    A roxygen2 bump and a package-level help file — the identical release MVMR shipped the same week, which is how tightly these two move together.

    View source ↗
  3. 4mo ago

    RMVMR 0.4.3

    Unspecified code optimizations, mirroring the same note in MVMR's release from the following day. No detail on scope or whether output is affected.

    View source ↗
  4. 5mo ago

    RMVMR now requires MVMR 0.4.3 or later

    A dependency floor and nothing else — pinned to the MVMR release that restored intercepts to the snpcov_mvmr() regressions. The clearest illustration that RMVMR's correctness is inherited rather than its own.

    View source ↗
  5. 1y ago

    RMVMR 0.4.1

    A scale_x_continuous() fix in plot_rmvmr() so the package keeps working under ggplot2 4. Forward-compatibility maintenance.

    View source ↗