sample_indices parameter missing from multiple methods that internally call sample_metadata

[noir4201]

Hello mentors,
While exploring the malariagen-data-python API in preparation for a GSoC 2026 proposal, I noticed a parameter inconsistency across several public methods.
Problem
sample_metadata() accepts a sample_indices parameter that allows users to select specific samples by their index position in the dataset. However, four other public methods — count_samples(), plot_samples_interactive_map(), plot_samples_bar() and plot_sample_location_geo() — all call sample_metadata() internally but do not expose sample_indices to the caller. This means a user who wants to work with a specific subset of samples selected by index can do so with sample_metadata() but cannot use the same selection with any of these four methods.
Evidence this is an oversight
wgs_data_catalog() is another method that calls sample_metadata() internally and it already correctly exposes and passes through sample_indices. This confirms the pattern exists and the four methods above were simply missed.
User impact
Users cannot count, plot for a sample subset selected by index, even though the underlying sample_metadata() fully supports it. There is no workaround within the existing API.
Proposed fix
Add sample_indices as an optional parameter to the signatures of all four affected methods, with a default value of None. Then pass it through to the internal sample_metadata() call in each method , exactly the same way wgs_data_catalog() already does it.

I'd be happy to work on a PR for this if it seems useful to contribute as part of my Gsoc proposal
Best regards,
noir4201

[noir4201]

HI Mentors,
I would like to correct a mistake in my original issue description, I incorrectly stated that wgs_data_catalog() correctly exposes and passes through sample_indices as evidence this was an oversight. On closer inspection, wgs_data_catalog() does not call sample_metadata() internally at all. The correct evidence is that the codebase has a consistent pattern where complex methods use _prep_sample_selection_cache_params() to handle sample_indices, while simpler methods should pass it directly through to sample_metadata(). The 4 methods fixed in PR #1127 were not following this pattern.

[khushthecoder]

Hi @jonbrenas,

This PR adds sample_indices support to count_samples(), plot_samples_interactive_map(), plot_samples_bar(), and plot_sample_location_geo().

These 4 methods call sample_metadata() internally but didn't expose its sample_indices parameter, so users couldn't count or plot a sample subset selected by index position. The fix is minimal — just forwarding the parameter through. No extra validation is needed since sample_metadata() already enforces the XOR constraint between sample_query and sample_indices.

Looking forward to your feedback!

I have submitted a PR addressing this issue. Could you please review it? #1159

[noir4201]

Hi @khushthecoder
Just wanted to flag that PR #1127 addresses the same issue and was submitted earlier. I also filed the original issue #1126. I have also checked your PR and it is exactly what I did. Happy to collaborate if needed feel free to check my implementation.

[khushthecoder]

Hi @noir4201,

Thank you for flagging this, and apologies for the overlap! I didn't see that you had already opened PR #1127 for your issue when I started working on it.

Since you are the original author of the issue and your PR was submitted first, it absolutely makes sense to move forward with yours. I'll go ahead and close this PR to avoid any confusion for the maintainers and keep the review process streamlined on your end.

Great catch on identifying this API inconsistency in the first place! Good luck with your GSoC 2026 proposal, and let me know if there's ever anything else I can collaborate with you on.

[noir4201]

Hi @khushthecoder
Thank you for being so gracious about this! Really appreciate it. Good luck with your GSoC application too , will be happy to collaborate anytime.