# How to use scib benchmark-metrics evaluate the adata after BBKNN integretion

**URL:** <https://discourse.scverse.org/t/how-to-use-scib-benchmark-metrics-evaluate-the-adata-after-bbknn-integretion/3807>\
**Category:** Help\
**Tags:** integration\
**Created:** [September 22, 2025, 3:38am UTC](https://discourse.scverse.org/t/how-to-use-scib-benchmark-metrics-evaluate-the-adata-after-bbknn-integretion/3807 "2025-09-22T03:38:00Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![wangflmi](https://yyz1.discourse-cdn.com/flex035/user_avatar/discourse.scverse.org/wangflmi/32/1496_2.png) [@wangflmi](https://discourse.scverse.org/u/wangflmi)\
**Post date:** [September 22, 2025, 3:38am UTC](https://discourse.scverse.org/t/how-to-use-scib-benchmark-metrics-evaluate-the-adata-after-bbknn-integretion/3807/1 "2025-09-22T03:38:00Z")

</div>

The paper showed we can use benchmark metrics to evaluate the adata after BBKNN integration. But I can only use some default functions in the benchmark-metrics package to get the batch-remove grades. For bio-conservation grades, I can’t find appropriate functions to apply.

---

<div class="post-metadata">

**Author:** ![ori-kron-wis](https://yyz1.discourse-cdn.com/flex035/user_avatar/discourse.scverse.org/ori-kron-wis/32/1099_2.png) [@ori-kron-wis](https://discourse.scverse.org/u/ori-kron-wis)\
**Post date:** [September 25, 2025, 12:11pm UTC](https://discourse.scverse.org/t/how-to-use-scib-benchmark-metrics-evaluate-the-adata-after-bbknn-integretion/3807/2 "2025-09-25T12:11:32Z")

</div>

Hey @wangflmi

Have you checked scib-metrics tutorials? There are several examples overthere on how to run scib-metrics benchmarker to produce batch-remove & bio-conservation grades

See for example:

> **[Benchmarking lung integration](https://scib-metrics.readthedocs.io/en/stable/notebooks/lung_example.html)**
>
> Here we walkthrough applying the integration benchmarking metrics on the lung atlas example from the scIB paper. Load and preprocess data: We subset to the highly variable genes so that each method...

---

<div class="post-metadata">

**Author:** ![cynthier](https://avatars.discourse-cdn.com/v4/letter/c/5f8ce5/32.png) [@cynthier](https://discourse.scverse.org/u/cynthier)\
**Post date:** [October 6, 2025, 2:57pm UTC](https://discourse.scverse.org/t/how-to-use-scib-benchmark-metrics-evaluate-the-adata-after-bbknn-integretion/3807/4 "2025-10-06T14:57:48Z")

</div>

Hi@ori-kron-wis,

The tutorials seemed only show the embedding-based methods, I could not find the example for the graph-based one, like BBKNN.

---

<div class="post-metadata">

**Author:** ![ori-kron-wis](https://yyz1.discourse-cdn.com/flex035/user_avatar/discourse.scverse.org/ori-kron-wis/32/1099_2.png) [@ori-kron-wis](https://discourse.scverse.org/u/ori-kron-wis)\
**Post date:** [October 7, 2025, 6:40am UTC](https://discourse.scverse.org/t/how-to-use-scib-benchmark-metrics-evaluate-the-adata-after-bbknn-integretion/3807/5 "2025-10-07T06:40:49Z")

</div>

Can you see if this helps you:

> <https://github.com/YosefLab/scib-metrics/issues/192>
>
> \### Description of feature
> 
> It would be great to show how to use precomputed nei…ghbors as input to Benchmarker. For example, when evaluating multiple cell type annotations for the same dataset, Benchmarker will recompute the kNN graph each time.
> 
> It's maybe already possible to do this manually by obtaining a NeighborsResults object and then applying individual metrics or by editing the Benchmarker attributes