Complete Model v2 wrapper implementation with shift parameters (#180)

* Complete Model v2 wrapper implementation with shift parameters (#172)

Implements comprehensive Model wrapper for jaxmodels backend with wide format
API, shift parameters, and extensive test coverage.

Key Features:
- Wide format get_mutations_df(): One row per mutation with beta_{condition}
  and shift_{condition} columns (matches v1 API)
- Shift parameters automatically calculated as β_condition - β_reference
- Extended Model.fit() to expose advanced jaxmodels parameters via kwargs
  (beta_clip_range, ge_kwargs, cal_kwargs, loss_kwargs)
- 24 comprehensive tests in test_model_v2.py covering all Model v2 functionality
- Updated jaxmodels_simulation_fits.ipynb to use Model wrapper API

Changes:
- multidms/model.py: Complete v2 Model wrapper implementation
- multidms/jaxmodels.py: Fixed type hints
- tests/test_model_v2.py: 24 comprehensive tests (NEW)
- tests/test_data.py: Commented out v1 Model tests
- notebooks/jaxmodels/jaxmodels_simulation_fits.ipynb: Updated to Model API

Tests: 61 passed, 1 skipped
All linting and formatting checks pass

Closes #172
Related: #173, #178, #179

* Fix CI: Exclude deprecated files from ruff linting

Add biophysical.py and model_v1_backup.py to ruff exclude list in pyproject.toml.
These are deprecated/backup files that should not block CI.

* Remove deprecated files from repository

Remove biophysical.py and model_v1_backup.py from version control as they
are no longer needed for v2 implementation. These were causing CI linting
failures and are superseded by the new jaxmodels-based Model wrapper.

Users can access v1 functionality by using multidms v1.x releases.

Changes:
- Removed multidms/biophysical.py (deprecated with ImportError)
- Removed multidms/model_v1_backup.py (backup of old implementation)
- Cleaned up pyproject.toml excludes (no longer needed)

* Optimize test_model.py for faster execution

- Replace test_model_v2.py with optimized test_model.py
- Add session-scoped data fixtures to avoid repeated creation
- Add module-scoped fitted model fixtures shared across tests
- Reduce iteration counts where sufficient (maxiter 5→3, 20→10)
- Expected ~3-4x speedup from fixture reuse and reduced iterations
- All 28 tests retained with 61 passed, 1 skipped

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

* Fix docs build errors in CI

- Fix CHANGELOG.rst link syntax (proper RST format)
- Fix jaxtyping array shape annotation (space not comma)

Fixes errors:
- "Unknown target name" in CHANGELOG.rst line 33
- "Axes should be separated with spaces, not commas" in jaxmodels.py

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

* remove unnecessary spec files, and simulated data

* remove old REFACTOR PLAN

* remove unnecessary claude commands

* update gitignore

---------

Co-authored-by: Claude <[email protected]>

Author: jgallowa07

.gitignore

@@ -7,6 +7,19 @@ Attic/
 docs/multidms*rst
 .vscode
 
+
+
+# ecperimental files
+notebooks/Smaller_beta_output/
+notebooks/linear_epistasis_output/
+notebooks/separate_offset_output/
+
+# Specify kit related
+.claude/
+.specify/
+REFACTOR_PLAN.md
+specs/
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]
@@ -171,4 +184,4 @@ cython_debug/
 
 notebooks/test_dump.pkl
 notebooks/jaxmodels/papermill_results
-notebooks/jaxmodels/output
+notebooks/jaxmodels/output

CHANGELOG.rst

@@ -30,7 +30,7 @@ The format is based on `Keep a Changelog <https://keepachangelog.com>`_.
 0.4.0
 -----
 - new simulation validation analysis and plotting functions (at the time of re-submission)
-- fixes bug described in `#130 https://github.com/matsengrp/multidms/issues/130`_, having to do with pandas groupby.apply 2.2.0 behavior change.
+- fixes bug described in `#130 <https://github.com/matsengrp/multidms/issues/130>`_, having to do with pandas groupby.apply 2.2.0 behavior change.
 - updates python version requirements to 3.9 or newer, as 3.8 did not work with the new pandas version, 2.2.0 bug patch described above.
 - supresses the cpu warning from jax.
 - adds `ModelCollection.add_validation_loss <https://github.com/matsengrp/multidms/blob/b0e7cbe96216e1307d070adc531fe51a960ec32a/multidms/model_collection.py#L569>`_, `ModelCollection.get_conditional_loss_df <https://github.com/matsengrp/multidms/blob/b0e7cbe96216e1307d070adc531fe51a960ec32a/multidms/model_collection.py#L627>`_, `Model.conditional_loss <https://github.com/matsengrp/multidms/blob/b0e7cbe96216e1307d070adc531fe51a960ec32a/multidms/model.py#L379>`_, and `Model.get_df_loss <https://github.com/matsengrp/multidms/blob/b0e7cbe96216e1307d070adc531fe51a960ec32a/multidms/model.py#L568>`_ methods, which can all be used quite easily to perform cross validation analysis.

CLAUDE.md

@@ -0,0 +1,92 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project Overview
+
+`multidms` is a Python package for joint modeling of multiple deep mutational scanning (DMS) experiments. It uses JAX for high-performance computing and automatic differentiation to fit global-epistasis models that estimate individual mutation effects and how they differ between experimental conditions.
+
+## Development Commands
+
+### Core Development Workflow
+```bash
+# Install development dependencies
+pip install -e ".[dev]"
+
+# Code quality checks
+ruff check .              # Lint code
+black .                   # Format code
+
+# Testing
+pytest --doctest-modules -vv    # Run all tests including doctests
+pytest tests/                   # Run unit tests only
+pytest --doctest-modules multidms tests  # Full test suite (as used in CI)
+
+# Documentation
+make -C docs clean        # Clean docs build
+make -C docs html         # Build documentation
+
+# Version management
+bumpver update --patch    # Bump patch version
+bumpver update --minor    # Bump minor version
+bumpver update --major    # Bump major version
+```
+
+### Testing Notes
+- The test suite is minimal with only basic Data class tests in `tests/test_data.py`
+- Doctests are integrated throughout the codebase and run with pytest
+- CI runs tests on Python 3.9, 3.10, 3.11 on Ubuntu and macOS
+
+## Package Architecture
+
+### Core Classes and Entry Points
+- **`multidms.Data`** - Handles data preprocessing and one-hot encoding of variant substitutions
+- **`multidms.Model`** - Main model class for fitting DMS experiments using JAX-based optimization
+- **`multidms.ModelCollection`** - Interface for fitting multiple models in parallel
+- **`multidms.fit_models`** - Function for parallel model fitting across collections
+
+### Key Modules
+- **`multidms.biophysical`** - Core biophysical model equations, transformations, and mathematical foundations
+- **`multidms.model_collection`** - Parallel model fitting and analysis workflows
+- **`multidms.plot`** - Interactive plotting functionality using matplotlib/seaborn/altair
+- **`multidms.utils`** - Data transformation utilities and helper functions
+
+### Dependencies and Architecture Patterns
+- **JAX ecosystem**: Core computational framework with jaxopt for optimization
+- **Data handling**: pandas for DataFrames, numpy for arrays (version pinned ≤1.26.0)
+- **Optimization**: Uses generalized lasso with bit-flipping algorithms via pylops/pyproximal
+- **Visualization**: Multi-library approach (matplotlib, seaborn, altair) for different plot types
+- **Scientific computing**: scipy for statistical functions, polyclonal for related modeling
+
+### Code Style and Conventions
+- **Formatting**: Black with line length 89 (matches ruff configuration)
+- **Linting**: Ruff with specific rule selections (E, F, UP, D) and custom ignores for docstring styles
+- **Documentation**: NumPy-style docstrings throughout
+- **Type hints**: Used where appropriate, with typing_extensions for compatibility
+
+### Development Patterns
+- Models compose biophysical equations from `multidms.biophysical` module
+- Heavy use of JAX transformations (jit, grad, vmap) for performance
+- Parameter initialization and transformation handled through dedicated methods
+- Cross-validation and simulation validation workflows built into model classes
+
+### File Organization
+- Main package code in `multidms/` with flat module structure
+- Jupyter notebooks in `notebooks/` demonstrate usage and validation
+- Sphinx documentation in `docs/` with linked notebook examples
+- Minimal test suite in `tests/` (expansion needed)
+
+### CI/CD and Release Process
+- GitHub Actions handle testing, linting, documentation builds
+- Automated PyPI publishing on tagged releases
+- Version management via bumpver tool with coordinated updates across files
+- Multi-platform testing ensures compatibility across development environments
+
+## Active Technologies
+- Python 3.9, 3.10, 3.11 (multi-version CI support) + JAX ≥0.4.29, jaxopt, equinox, pandas ≥2.2.0, numpy ≤1.26.0 (001-jaxmodels-refactor)
+- N/A (library operates on in-memory pandas DataFrames) (001-jaxmodels-refactor)
+- Python 3.9, 3.10, 3.11 (CI tested on all three versions) + JAX ≥0.4.29, equinox, jaxopt, pandas ≥2.2.0, numpy ≤1.26.0 (001-jaxmodels-refactor)
+- N/A (in-memory DataFrame processing) (001-jaxmodels-refactor)
+
+## Recent Changes
+- 001-jaxmodels-refactor: Added Python 3.9, 3.10, 3.11 (multi-version CI support) + JAX ≥0.4.29, jaxopt, equinox, pandas ≥2.2.0, numpy ≤1.26.0