[ENH] aptacom database loader #158

rpgv · 2025-09-25T09:10:56Z

Added: AptaCom loader - Loads AptaCom dataset from hugging face in one of two formats:
pandas dataframe, hugging face dataset;

…emistry. Both applied in a broad manner

fkiraly · 2025-09-29T20:00:49Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+
+    Parameters
+    ----------
+    as_df: (bool) Requires pandas compatible format; converts


defaults should be stated in the docstring, and please adopt a consistent formatting. Suggestion:

paramname : type/description, default=default_value

This is what we use.

fkiraly · 2025-09-29T20:01:17Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+
+    Returns
+    -------
+    dataset: Hugging face dataset in a pandas compatible format.


I suggest to be more precise here, and also take into account the value of as_df, e.g., what is returned if it is False (the default actually)?

fkiraly · 2025-09-29T20:01:52Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+    as_df: (bool) Requires pandas compatible format; converts
+    dataset into pandas dataframe
+
+    filter(str): Allows filtering dataset by:


parameter is called differently in function signature

fkiraly

Very nice!

I had comments mainly on the docstring - AI is also good (under supervision) in writing clear docstrings.

It would also be great if a description of the dataset, e.g., which columns it has, a reference to a paper, etc, would be part of the docstring.

fkiraly · 2025-09-29T20:04:26Z

pyaptamer/datasets/tests/test_aptacom_loader.py

+    The test_download_aptacom function
+    """
+    dataset = load_aptacom(as_df, filter_entries)
+    if not isinstance(dataset, Dataset | DataFrame):


I would actually test the correct type based on as_df.

Also, can you check number rows, expected columns, etc?

- load_aptacom function renamed to load_complete_aptacom; - Added 'load_aptacom_for_training' function which returns a dataset or dataframe with aptamer and target sequence, with no metadata; - Both functions within _aptacom_loader.py allow filtering the AptaCom dataset through target or aptamer chemistry;

fkiraly · 2025-10-08T22:03:37Z

code quality tests are failing - I would recommend to set up pre-commit locally, that way code is always formatted correctly (automatically)

fkiraly

Looks good now!

satvshr · 2025-10-10T23:43:20Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+
+
+def load_complete_aptacom(as_df=False, filter_entries=None):
+    """Loads a Hugging Face dataset with customizable


Can we replace "a Hugging Face" dataset everywhere in the docstring with the actual thing it is loading? Something like: "the AptaCom dataset from HuggingFace"

Of course - I'll modify it then.

satvshr · 2025-10-10T23:52:40Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+                ]
+            )
+        else:
+            aptacom = aptacom.map(


Given the only difference between load_complete_aptacom and load_aptacom_for_training seem to be the tagret_id argument, when the tagret_id is false, does load_aptacom_for_training give the same result as load_complete_aptacom?

No, 'load_for_training' doesn't contain any 'metadata' relative to the aptamer entries. It only describes aptamer sequence, target sequence (and if selected, target id, which is required for AptaCom model).

If load_aptacom_for_training returns data that is a subset of load_complete_aptacom can we not just add 2 arguments to load_complete_aptacom (something like load_training and target_id) and remove the load_aptacom_for_training function altogether?

fkiraly

After discussion some changes:

from @NennoMP: for loading, we may be able to reuse the huggingface loader
since you are subsetting after loading, in load_for_train", I would instead call the load_complete`, so the loader code is not duplicated
I would not merge these into a single function, since these correspond to two "data artefacts" that may be worth keeping separate. Artefacts should map on python objects imo.

minor things:

naming; I would say load_aptacom_full, and load_aptacom_Xy - more descriptive of content, not purpose
- having affinity might be useful, so the second could return an X (aptamer, ligand sequence) and an y (affinity)
- this could use the return_X_y logic like in sklearn, and X with columns "target" and "ligand"
docstrings should be in numpydoc format, not google format

Renamed load_aptacom_complete -> load_aptacom_full; Created function load_aptacom_xy -> Following sklearn format; Novel conditions for filtering load_aptacom_full;

fkiraly

Looks good to me!

satvshr · 2025-10-26T19:32:17Z

Will have a look at it tomorrow or day after.

satvshr · 2025-10-27T20:45:24Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+
+
+def filter_columns(ds, columns=None):
+    """ " Selects columns to keep on dataset


I do not understand the """ " formatting everywhere, seems wrong?

satvshr · 2025-10-27T20:47:27Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+    """ " Selects columns to keep on dataset
+    Parameters:
+    -----------
+        ds: pd dataframe, required


Formatting seems a bit off, can you check this?

I'm sorry for the delayed response @satvshr, thank you for your comments - I'll make the adjustments !

Check #192 :D

satvshr · 2025-10-27T20:48:53Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+    return ds
+
+
+def prepare_xy(ds):


@fkiraly just a question, should we not name all functions with x and y in their name similarly? So this would be prepare_x_y instead.

satvshr · 2025-10-27T20:50:49Z

pyaptamer/datasets/_loaders/_aptacom_loader.py

+}
+
+
+def filter_columns(ds, columns=None):


Should this function not be private? Also do you mind changing ds to df (for dataframe) given ds is something I will be using for the Dataset object from HuggingFace?

satvshr

@fkiraly we will be replacing the load_hf_dataset function with the one I am working on in #155, so should this be paused till then so that it can be updated with the Hugging Face loader we will be using?

fkiraly · 2025-10-28T23:19:12Z

so should this be paused till then so that it can be updated with the Hugging Face loader we will be using?

I do not think we should block this. I would instead refactor, keeping the public interfaces the same.

satvshr · 2025-10-28T23:22:52Z

so should this be paused till then so that it can be updated with the Hugging Face loader we will be using?

I do not think we should block this. I would instead refactor, keeping the public interfaces the same.

I left some comments, could you have a look and tell me why they are not something that should be changed so that I do not raise such points again?

fkiraly · 2025-11-01T14:03:25Z

@satvshr, apologies, these were very valid points - I overlooked them somehow. Must have had a cached version of the PR displayed without your comments?

Rafael Vieira and others added 19 commits September 24, 2025 16:37

Created simplest aptacom hf integration

d716bc1

Added format parameters to load_aptacom

00761a2

Corrected format parameter error on load_aptacom

0f86c18

Corrected format load_aptacom

a175f9c

---

8a7a4a1

---

4fed192

Added test_aptacom_loader file, added dependencies to pyproject

6ac85f7

Modified formatting options

7ebdb1c

Removed format_ variable

15cd9b3

Corrected test_aptacom_loader

86cd6b5

---

443624e

---

dbd1912

Pre-commit format

b57655e

Sorted __init__ imports

7eeafdf

Sorted __init__ imports

e2d2b72

Sorted _loaders/__init__ imports

056a644

rm gn dir from data

b0b6716

Added option to filter AptaCom based on: aptamer chemistry, target ch…

497dcbf

…emistry. Both applied in a broad manner

-

aee137f

fkiraly added the enhancement New feature or request label Sep 29, 2025

fkiraly reviewed Sep 29, 2025

View reviewed changes

fkiraly requested changes Sep 29, 2025

View reviewed changes

fkiraly changed the title ~~Integrate aptacom database loader~~ [ENH] aptacom database loader Sep 29, 2025

fkiraly reviewed Sep 29, 2025

View reviewed changes

fkiraly assigned rpgv Sep 30, 2025

Rafael added 2 commits October 6, 2025 18:47

Added type verification to boolean arguments

19dd6c8

Reverified pre-commit modifications

ad31e46

fkiraly approved these changes Oct 10, 2025

View reviewed changes

Update pyproject.toml

36c2406

satvshr reviewed Oct 10, 2025

View reviewed changes

Merge branch 'main' into integrate-aptacom-database

e81499b

fkiraly requested changes Oct 14, 2025

View reviewed changes

Rafael added 4 commits October 17, 2025 14:25

Implemented _hf_loader into _aptacom_loader;

74586a8

Renamed load_aptacom_complete -> load_aptacom_full; Created function load_aptacom_xy -> Following sklearn format; Novel conditions for filtering load_aptacom_full;

-

4e828cc

Updated docstring

53e9360

pre commit format

0555663

rpgv requested a review from fkiraly October 22, 2025 08:17

fkiraly approved these changes Oct 26, 2025

View reviewed changes

satvshr reviewed Oct 27, 2025

View reviewed changes

fkiraly merged commit 95ac1b1 into main Oct 28, 2025
15 checks passed

fkiraly mentioned this pull request Nov 1, 2025

[DOC] improve aptacom loader docstrings #192

Open



		def load_complete_aptacom(as_df=False, filter_entries=None):
		"""Loads a Hugging Face dataset with customizable



		def filter_columns(ds, columns=None):
		""" " Selects columns to keep on dataset

[ENH] aptacom database loader #158

[ENH] aptacom database loader #158

Uh oh!

Conversation

rpgv commented Sep 25, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fkiraly left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fkiraly commented Oct 8, 2025

Uh oh!

fkiraly left a comment

Choose a reason for hiding this comment

Uh oh!

satvshr Oct 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fkiraly left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fkiraly left a comment

Choose a reason for hiding this comment

Uh oh!

satvshr commented Oct 26, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

satvshr Oct 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

satvshr left a comment

Choose a reason for hiding this comment

Uh oh!

fkiraly commented Oct 28, 2025

Uh oh!

Uh oh!

satvshr commented Oct 28, 2025

Uh oh!

fkiraly commented Nov 1, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

satvshr Oct 10, 2025 •

edited

Loading

fkiraly left a comment •

edited

Loading

satvshr Oct 27, 2025 •

edited

Loading