# Benchmarking CellHint with scib-metrics Benchmarker tool how to use embedding\_obsm\_keys

**URL:** <https://discourse.scverse.org/t/benchmarking-cellhint-with-scib-metrics-benchmarker-tool-how-to-use-embedding-obsm-keys/2216>\
**Category:** Help\
**Tags:** scrna-seq, integration, scanvi, scvi\
**Created:** [April 12, 2024, 2:40pm UTC](https://discourse.scverse.org/t/benchmarking-cellhint-with-scib-metrics-benchmarker-tool-how-to-use-embedding-obsm-keys/2216 "2024-04-12T14:40:11Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![bocchiv](https://avatars.discourse-cdn.com/v4/letter/b/ad7895/32.png) [@bocchiv](https://discourse.scverse.org/u/bocchiv)\
**Post date:** [April 12, 2024, 2:40pm UTC](https://discourse.scverse.org/t/benchmarking-cellhint-with-scib-metrics-benchmarker-tool-how-to-use-embedding-obsm-keys/2216/1 "2024-04-12T14:40:11Z")

</div>

Hi!  
I am integrating different single-cell atlases and testing different methods like scVI and scANVI and then benchamrking them using scib-metrics package to assess the quality of the integration. However, I also wanted to integrate the newly developed mthod called CellHint [Google Colab](https://colab.research.google.com/github/Teichlab/cellhint/blob/main/docs/notebook/cellhint_tutorial_integration.ipynb). However CellHint calculated the new corrected distances and connectivies that are then used to draw the new batch corrected umap. I was wondering if I can use the CellHint umap embedding in embedding\_obsm\_keys and compare them to the scVI and scANVI embeddings. Thank you for your help!

---

<div class="post-metadata">

**Author:** ![cane11](https://yyz1.discourse-cdn.com/flex035/user_avatar/discourse.scverse.org/cane11/32/241_2.png) [@cane11](https://discourse.scverse.org/u/cane11)\
**Post date:** [April 25, 2024, 9:27pm UTC](https://discourse.scverse.org/t/benchmarking-cellhint-with-scib-metrics-benchmarker-tool-how-to-use-embedding-obsm-keys/2216/2 "2024-04-25T21:27:48Z")

</div>

This is the general way to perform the comparison. scib-metrics does not work with precomputed neighborhood graphs and requires an embedding.