feat: add generic Manhattan/QQ plotting helpers for association results

[Sharon-codes]

Summary

Adds Manhattan and QQ plotting helpers to the API for association-style results, designed to work directly in Python workflows (including GWAS notebooks) without requiring R or Dash-specific tooling.

Design choices (addressing issue discussion)

• Not tied to GWAS only: the functions operate on generic association result tables containing contig/position/p-values.
• Python-native API support: no R dependency.
• No Dash requirement: uses Plotly, already used in this project.
• Supports multiple data backends:
  • pandas.DataFrame (e.g. PLINK-exported tables loaded in pandas)
  • xarray.Dataset (aligned with sgkit-style workflows)

Changes

• malariagen_data/anoph/phenotypes.py
  • Added plot_manhattan(...)
  • Added plot_qq(...)
  • Added internal normalization helper to accept both DataFrame and xarray Dataset inputs.
  • Added Manhattan controls for contig_order, contig_spacing, and threshold line.
• tests/anoph/test_phenotypes_plotting.py
  • Added tests for DataFrame and xarray inputs
  • Added validation test for missing columns

Validation

• poetry run pytest tests/anoph/test_phenotypes_plotting.py -q (5 passed)
• poetry run mypy malariagen_data/anoph/phenotypes.py tests/anoph/test_phenotypes_plotting.py --ignore-missing-imports
• poetry run ruff check malariagen_data/anoph/phenotypes.py tests/anoph/test_phenotypes_plotting.py

Closes #772

[Sharon-codes]

@jonbrenas Hey please look into this , i hope this addresses your question also which was there in the comment of the issue , if you need me to add anything do let me or else please check and allow it , thanks alot !!!

[jonbrenas]

Thanks @Sharon-codes! I don't think calling the module containing the code for the Manhattan and QQ plots 'phenotypes' makes a lot of sense (but feel free to prove me wrong). I don't think there is (currently) any function that would generate the type of input that these functions need, which reduces their usefulness quite significantly. It would also be great if the tests followed the same general approach that is used for other modules.

[Sharon-codes]

Thanks @jonbrenas — I’ve reworked this based on your comments. (totally agreed on your comment, lol wont be able to prove you wrong)

Changes made:

• Moved generic Manhattan/QQ plotting code out of anoph/phenotypes.py into a dedicated module: anoph/association.py.
• Added snp_phenotype_association() / association_results() so there is now a built-in way to generate plot input (contig, position, pvalue) from phenotypes_with_snps().
• Wired the new association mixin into AnophelesDataResource.
• Reworked tests and replaced test_phenotypes_plotting.py with tests/anoph/test_association.py.

Please review commit 148c84c.

[Sharon-codes]

hey @jonbrenas please accept this pr , thanks !!!

[jonbrenas]

Thanks @Sharon-codes. This looks great! Could you create a short notebook (with dummy phenotypic data as we are short on real data) showing how the functions are used, what the plots look like, ...

[Sharon-codes]

@jonbrenas Heyyyy added , please check !! Thanks for the input and close the PR if happy with the new file !!

[Sharon-codes]

Heyy @jonbrenas If everything is as per your liking , please close this PR thanksss

[Sharon-codes]

Heyyy @jonbrenas added all the changes as you asked , if all good , please accept and close the PR , thanksss !!

[Sharon-codes]

Heyyy @jonbrenas I have created the notebook as requested , please check and if all good , close the PR , thanksss !!

[jonbrenas]

Thanks @Sharon-codes. Would it be possible to create the phenotypic data as part of conftest.py (even if it is for only one of the resources (probably Ag3) and for a single sample set) so that the tests can follow the normal pattern?

[Sharon-codes]

Heyyyy @jonbrenas

Implemented this in commit f72b8c6.

• Added simulated phenotype data generation in tests/anoph/conftest.py for Ag3 (single sample set: AG1000G-AO).
• Updated tests/anoph/test_association.py to use the normal fixture-driven API pattern (Ag3 simulator fixture) instead of a standalone dummy API.

This keeps the association tests aligned with the existing anoph test structure.

If this sounds good , please merge and close the PR !!

[Sharon-codes]

heyyy @jonbrenas i have added the changes you asked for , if everything seems good , please close the PR , thanksss !!

[Sharon-codes]

Heyy @jonbrenas if all good , please accept it , thanks

[jonbrenas]

Quite a few of these should be defined as params in a separate file for consistency, maybe base_params.field is enough.