Skip to content

Releases

larmorx 0.1.0rc1

larmorx 0.1.0rc1 Pre-release
Pre-release

Choose a tag to compare

@github-actions github-actions released this 11 Nov 06:43

The first release, as a release candidate: a pre-release, published to check that the packages install and work on every platform before 0.1.0. LARMORX is in early development: the tools below are complete and compared case by case with the original programs, the pipeline larmorprepx is not ready yet, and FreeSurfer's tools are still to come.

What is in it

  • Three packages on PyPI, at one version:
    • larmorx (Apache-2.0): the Python library (import larmorx as lx), the commands larmorx and lx, which take the original programs' arguments (larmorx ants antsRegistration ...), and larmorprepx (a preview, below). It holds the ports of ANTs 2.6.5 and ITK 5.4.5, of sdcflows 2.17.1 with niworkflows' helpers and fMRIPrep's resampler and field-map steps; its own NIfTI-1/2 reading and writing (compared with nibabel's); the clean-room originals of AFNI's and FSL's tools; and the exact replicas of AFNI 26.3.05's tools (with CC BY 4.0's attribution).
    • larmorx-nc (the FSL Licence: non-commercial use only): exact replicas of FSL 6.0.7.23's mcflirt, BET (bet, bet2, betsurf, standard_space_roi, fslroi), FLIRT and its programs, FAST, topup and applytopup, fslmaths, fslstats and prelude, with the FSL files they read (three standard images, topup's settings files).
    • larmorx-gpl (the GNU GPL version 3): the exact replica of tedana 26.0.3's t2smap.
    • The extras: larmorx[prep] adds what larmorprepx will need (templateflow, matplotlib, nilearn, pandas, jinja2); larmorx[all] adds [prep] and the two companions at the same version.
  • Wheels for six platforms: Linux x86-64 and aarch64 (glibc 2.28 or newer), macOS 11 or newer (Intel and Apple silicon) and Windows 10 and 11 (x64 and arm64; the C runtime is built in, no Visual C++ installation needed). One wheel per platform serves Python 3.12 and every newer version (CPython's stable ABI); tested on 3.12 on every platform, and on 3.13 and 3.14 on Linux. The source distributions of the three packages build on other systems with Rust and a C compiler.
  • Standalone programs on the GitHub Release, for each platform: larmorx (with lx), larmorx-nc and larmorx-gpl as archives that need no Python, each with its licences (larmorx-nc's with FSL's original and amended source, as its licence requires).
  • Two container images for linux/amd64 and linux/arm64, usable by Apptainer, Docker and Podman: ghcr.io/karellopez/larmorx:0.1.0rc1 (larmorx with [prep], without the replica packages: commercial use allowed) and ghcr.io/karellopez/larmorx:0.1.0rc1-all (also larmorx-gpl and larmorx-nc: non-commercial use only).
  • Checks for every file: its SHA-256 in the release's SHA256SUMS, and its build provenance (GitHub's attestations, PyPI's for its uploads).
  • larmorx --version, larmorx-nc --version and larmorx-gpl --version print the version as PyPI writes it (0.1.0rc1).

The tools and their status

Every tool has a validation record that compares it with the original program, case by case (the validation index; the numbers are in the README's tool table). A record's status means:

  • validated: every case of its comparison is within its thresholds (bit-identical, for an exact replica);
  • experimental: the tool runs and is compared with the original, but not every case is within its thresholds yet, or its comparison is not complete. A tool can be experimental although every compared case agrees: its record says what it has been compared on. The final validation phase (the fixed data panel, the full test tiers, long random-testing runs), still to come, settles the rest.

A record's name ending in -replica is the exact replica's (in larmorx itself for AFNI's tools, in larmorx-nc for FSL's, in larmorx-gpl for tedana's); -full is a record's full tier; the other records are the port's or the clean-room original's.

Tool Validation records and their status
NIfTI-1/2 I/O nifti-io: validated
antsApplyTransforms ants-apply-transforms: validated
ITK transform files (.mat, .txt, .h5) itk-hdf5: validated
N4BiasFieldCorrection ants-n4: validated
DenoiseImage ants-denoise: validated
ImageMath (every operation fMRIPrep, sMRIPrep and sdcflows call) ants-image-math: validated
ThresholdImage, MultiplyImages, SmoothImage, ResampleImage, ResampleImageBySpacing ants-threshold-image: validated; ants-multiply-images: validated; ants-smooth-image: validated; ants-resample-image: validated; ants-resample-image-by-spacing: validated
antsRegistration (every stage: linear, SyN, B-spline SyN, B-spline, displacement-field, exponential and time-varying velocity-field; image and point-set metrics) ants-registration: validated; ants-registration-corrections: validated
antsAI, antsAffineInitializer ants-ai: validated; ants-affine-initializer: validated
Atropos (every initialization, likelihood model, MRF, outlier and partial-volume option) ants-atropos: validated
fMRIPrep's one-shot BOLD resampler resample-series: validated
sdcflows' field-map estimation (the B-spline field fit, the field evaluated and applied, the phase-difference preparation) sdcflows-fieldmap: validated
sdcflows' PEPOLAR, direct and fieldmap-less estimation (the masks, the estimators of a BIDS dataset, the derivatives, the BIDS app; niworkflows' helpers; fMRIPrep's own field-map steps) sdcflows-estimation: validated; sdcflows-estimation-full: validated; sdcflows-estimation-app-shared: validated; niworkflows-helpers: validated; ndimage-skimage: validated; ndimage-skimage-full: validated
tedana's t2smap (the T2* and S0 maps of multi-echo fMRI and the echoes' optimal combination) tedana-t2smap-replica: experimental
3dTshift (slice timing) afni-tshift: validated; afni-tshift-replica: validated
3dAutomask (brain mask) afni-automask: validated; afni-automask-replica: validated
3dUnifize (intensity uniformization) afni-unifize: validated; afni-unifize-replica: validated
3dvolreg (rigid-body registration of a series) afni-volreg-original: experimental; afni-volreg-replica: validated
Head-motion correction (mcflirt) mri-hmc: experimental; fsl-mcflirt-replica: validated; fsl-mcflirt-replica-full: validated
BET (brain extraction; the skull and scalp surfaces) [fsl-bet](https://github...
Read more