[us-commits] [ehb54/ultrascan3] 9ecb32: extrap_c0: fix Guinier-QC window (point-count -> q...

[emre brookes]

  Branch: refs/heads/ehb54-issue-920
  Home:   https://github.com/ehb54/ultrascan3
  Commit: 9ecb32097ef10d9de0f1d835f0d6319b77527b1d
      https://github.com/ehb54/ultrascan3/commit/9ecb32097ef10d9de0f1d835f0d6319b77527b1d
  Author: ehb54 <brookes at uthscsa.edu>
  Date:   2026-07-16 (Thu, 16 Jul 2026)

  Changed paths:
    M us_somo/develop/src/us_hydrodyn_saxs_iqq_extrap_c0.cpp

  Log Message:
  -----------
  extrap_c0: fix Guinier-QC window (point-count -> q-value) so fine grids work

The standalone Guinier QC of the extrapolated curve (us_extrap_c0_guinier, a
rough automatic ln I vs q^2 check, NOT SOMO's Guinier panel) chose its initial
low-q window by POINT COUNT (first 25 points). On a coarse grid that is ~q<0.08;
on a fine, densely sampled grid (e.g. the 2452-point SASBDB deposits, q from
0.0032) 25 points spans only q<0.008 -- a tiny, noise-dominated sliver -- so the
fit returned a nonsensical Rg (86 A on urate oxidase) or no fit at all (RNaseA,
xylanase), even though the extrapolated CURVE was correct.

Choose the initial window by q-VALUE (q <= 0.08 A^-1, grid-density independent,
matching the Python reference and SOMO's own q*Rg<=1.3 convention), fall back to
a point count only for coarse/short grids, and iterate the q*Rg<=1.3 refinement
to convergence instead of a single pass. Also label the log line as a rough
automatic check and point to the Guinier panel for a definitive fit.

Validated on the SASBDB Table-1 additive benchmark: urate oxidase 86 -> 37.0,
RNaseA/xylanase no-fit -> 15.1 / 15.0, glucose isomerase 32.7, lysozyme 14.1 --
all matching a reference Guinier run on the same output curves. Only the QC
readout changed; the extrapolated curves are unchanged.

Co-Authored-By: Claude Fable 5 <noreply at anthropic.com>



To unsubscribe from these emails, change your notification settings at https://github.com/ehb54/ultrascan3/settings/notifications