# \#developer

**URL:** https://discourse.scverse.org/tag/developer/24.md

[Latest](https://discourse.scverse.org/latest.md) · [Categories](https://discourse.scverse.org/categories.md) · [Tags](https://discourse.scverse.org/tags.md)

---

## [Squidpy calculations on giant data](https://discourse.scverse.org/t/squidpy-calculations-on-giant-data/4041)

<div class="topic-metadata">

**Author:** [@Sudden\_Deceleration](https://discourse.scverse.org/u/Sudden_Deceleration)\
**Replies:** 0\
**Last updated:** [August 4, 2026, 11:16am UTC](https://discourse.scverse.org/t/squidpy-calculations-on-giant-data/4041 "2026-08-04T11:16:59Z")

</div>

I only today started looking into spatial analysis and squidpy, so please be patient with me if this has an easy answer. I loaded my data from a whole slide image with 754130 cells, each one carrying a cell type and coo…

---

## [metal-SingleCell — a work-in-progress Metal/MLX port of rapids-singlecell for Apple Silicon (feedback welcome)](https://discourse.scverse.org/t/metal-singlecell-a-work-in-progress-metal-mlx-port-of-rapids-singlecell-for-apple-silicon-feedback-welcome/4018)

<div class="topic-metadata">

**Author:** [@gingerii](https://discourse.scverse.org/u/gingerii)\
**Replies:** 1\
**Last updated:** [July 15, 2026, 6:48am UTC](https://discourse.scverse.org/t/metal-singlecell-a-work-in-progress-metal-mlx-port-of-rapids-singlecell-for-apple-silicon-feedback-welcome/4018 "2026-07-15T06:48:21Z")

</div>

TL;DR: I’ve been working on an independent Metal/MLX re-implementation of rapids-singlecell so the scanpy/squidpy GPU workflow can run on the Apple M-series GPU. It’s a drop-in API and covers ~30 pp/tl/gr functions, vali…

---

## [Macbook M1 M2 mps acceleration with scVI](https://discourse.scverse.org/t/macbook-m1-m2-mps-acceleration-with-scvi/2075)

<div class="topic-metadata">

**Author:** [@davemcg](https://discourse.scverse.org/u/davemcg)\
**Replies:** 10\
**Last updated:** [March 3, 2026, 11:43am UTC](https://discourse.scverse.org/t/macbook-m1-m2-mps-acceleration-with-scvi/2075 "2026-03-03T11:43:54Z")

</div>

Has anyone recently gotten scVI (ideally 1.0.4) working with “GPU” (well, “mps”) acceleration with a Apple ARM M1, M2, or M3? I’ve tried a variety of incantations when installing torch and jax and it either doesn’t see t…

---

## [Sc.pp.filter\_genes how to use](https://discourse.scverse.org/t/sc-pp-filter-genes-how-to-use/3901)

<div class="topic-metadata">

**Author:** [@baicai](https://discourse.scverse.org/u/baicai)\
**Replies:** 0\
**Last updated:** [December 16, 2025, 6:41am UTC](https://discourse.scverse.org/t/sc-pp-filter-genes-how-to-use/3901 "2025-12-16T06:41:06Z")

</div>

I would like to ask for clarification regarding the recommended order of sc.pp.filter\_genes and sc.pp.normalize\_total, and the intended interpretation of their interaction. It is clear that filtering genes (e.g. min\_c…

---

## [Inquiry Regarding DRVI Model Compatibility with scANVI for Embedding Optimization?](https://discourse.scverse.org/t/inquiry-regarding-drvi-model-compatibility-with-scanvi-for-embedding-optimization/3876)

<div class="topic-metadata">

**Author:** [@Xicici-Yan](https://discourse.scverse.org/u/Xicici-Yan)\
**Replies:** 1\
**Last updated:** [November 25, 2025, 7:21pm UTC](https://discourse.scverse.org/t/inquiry-regarding-drvi-model-compatibility-with-scanvi-for-embedding-optimization/3876 "2025-11-25T19:21:12Z")

</div>

I am writing to inquire about the compatibility of the DRVI model with scANVI for embedding optimization. Specifically, I would like to know: Whether the DRVI model supports using scANVI to optimize embeddings, and if…

---

## [Help with package setup](https://discourse.scverse.org/t/help-with-package-setup/3811)

<div class="topic-metadata">

**Author:** [@josenimo](https://discourse.scverse.org/u/josenimo)\
**Replies:** 2\
**Last updated:** [October 7, 2025, 10:46am UTC](https://discourse.scverse.org/t/help-with-package-setup/3811 "2025-10-07T10:46:39Z")

</div>

Hey all, First, I am not sure this is the right place to post this. But I looked through this discourse and through Zulip, and I could not find a better place. I am a single developer working on the openDVP package, wh…

---

## [Domain adaptation to pre-train batch correction model using paired data](https://discourse.scverse.org/t/domain-adaptation-to-pre-train-batch-correction-model-using-paired-data/3633)

<div class="topic-metadata">

**Author:** [@Valerie](https://discourse.scverse.org/u/Valerie)\
**Replies:** 12\
**Last updated:** [May 27, 2025, 6:44am UTC](https://discourse.scverse.org/t/domain-adaptation-to-pre-train-batch-correction-model-using-paired-data/3633 "2025-05-27T06:44:23Z")

</div>

Good afternoon scvi team! I am very new to the scvi tool and moslty have been working with scRNA-seq data within R environment. I came around a specific issue which scvi could potentially be well suited for so I would l…

---

## [Daf (Data in Axes in Formats)](https://discourse.scverse.org/t/daf-data-in-axes-in-formats/2531)

<div class="topic-metadata">

**Author:** [@orenbenkiki](https://discourse.scverse.org/u/orenbenkiki)\
**Replies:** 0\
**Last updated:** [September 24, 2024, 6:01am UTC](https://discourse.scverse.org/t/daf-data-in-axes-in-formats/2531 "2024-09-24T06:01:23Z")

</div>

I understand there are ongoing discussions about the future of AnnData. I have just released alpha versions of GitHub - tanaylab/DataAxesFormats.jl: Data in Axes in Formats and dafpy · PyPI - R wrappers are WIP. This t…

---

## [Access to each training example's original index in the input Anndata object](https://discourse.scverse.org/t/access-to-each-training-examples-original-index-in-the-input-anndata-object/2153)

<div class="topic-metadata">

**Author:** [@g12bftd](https://discourse.scverse.org/u/g12bftd)\
**Replies:** 0\
**Last updated:** [March 12, 2024, 4:18pm UTC](https://discourse.scverse.org/t/access-to-each-training-examples-original-index-in-the-input-anndata-object/2153 "2024-03-12T16:18:01Z")

</div>

I am training scvi with augmentations. For one of my augmentations, I need access to each cell’s original index in the full input adata object. What is the best way to go about this?

---

## [Batch correction using scvI on multiple datasets + hyperparameter tuning of an scvI model](https://discourse.scverse.org/t/batch-correction-using-scvi-on-multiple-datasets-hyperparameter-tuning-of-an-scvi-model/2076)

<div class="topic-metadata">

**Author:** [@g12bftd](https://discourse.scverse.org/u/g12bftd)\
**Replies:** 1\
**Last updated:** [February 16, 2024, 5:45pm UTC](https://discourse.scverse.org/t/batch-correction-using-scvi-on-multiple-datasets-hyperparameter-tuning-of-an-scvi-model/2076 "2024-02-16T17:45:21Z")

</div>

I want to perform batch integration on multiple datasets (loaded as separate h5ad files). Is there an exemplar code snippet on how to do this? Once I’ve loaded and pre-processed the data, what is the best way to rigorou…

---

## [Trying to read custom genome using sc.read\_10x\_h5](https://discourse.scverse.org/t/trying-to-read-custom-genome-using-sc-read-10x-h5/2029)

<div class="topic-metadata">

**Author:** [@mgelm](https://discourse.scverse.org/u/mgelm)\
**Replies:** 0\
**Last updated:** [January 26, 2024, 12:51pm UTC](https://discourse.scverse.org/t/trying-to-read-custom-genome-using-sc-read-10x-h5/2029 "2024-01-26T12:51:51Z")

</div>

Hello! I’m having trouble reading my .h5 files to a 10x mm10 Custom Genome containing LacZ. When I run this file in Seurat it picks up the LacZ gene but in scanpy the gene seems to be missing. I tried to run the convert…

---

## [Questions about scib weight choice](https://discourse.scverse.org/t/questions-about-scib-weight-choice/1941)

<div class="topic-metadata">

**Author:** [@HelloWorldLTY](https://discourse.scverse.org/u/HelloWorldLTY)\
**Replies:** 0\
**Last updated:** [December 5, 2023, 4:16am UTC](https://discourse.scverse.org/t/questions-about-scib-weight-choice/1941 "2023-12-05T04:16:37Z")

</div>

Hi, I have a quick question about the setting of the weighted sum in scIB: I understand to assgin S\_bio with 0.6 and S\_batch as 0.4 are to ensure bio convservation is more important. However, I wonder what is the mot…

---

## [UserWarning: No data for colormapping provided via 'c'. Parameters 'cmap' will be ignored](https://discourse.scverse.org/t/userwarning-no-data-for-colormapping-provided-via-c-parameters-cmap-will-be-ignored/1221)

<div class="topic-metadata">

**Author:** [@DRSEI](https://discourse.scverse.org/u/DRSEI)\
**Replies:** 8\
**Last updated:** [December 1, 2023, 4:36pm UTC](https://discourse.scverse.org/t/userwarning-no-data-for-colormapping-provided-via-c-parameters-cmap-will-be-ignored/1221 "2023-12-01T16:36:21Z")

</div>

Hello, I am experiencing an issue with my UMAP plots in Scanpy. When trying to plot my data, the process takes over 20 minutes and does not produce any output. I have updated my Scanpy to the latest version and re-insta…

---

## [Expression\_cutoff optiion from scanpy.pl.dotplot function](https://discourse.scverse.org/t/expression-cutoff-optiion-from-scanpy-pl-dotplot-function/1786)

<div class="topic-metadata">

**Author:** [@fadhlyemen](https://discourse.scverse.org/u/fadhlyemen)\
**Replies:** 0\
**Last updated:** [September 28, 2023, 5:05pm UTC](https://discourse.scverse.org/t/expression-cutoff-optiion-from-scanpy-pl-dotplot-function/1786 "2023-09-28T17:05:16Z")

</div>

I have a question regarding the expression\_cutoff option from scanpy.pl.dotplot function. when and when not I should use this threshold? I got different results when I set a threshold. Thank you

---

## [How to best integrate Genomic Variant Information](https://discourse.scverse.org/t/how-to-best-integrate-genomic-variant-information/1673)

<div class="topic-metadata">

**Author:** [@votti](https://discourse.scverse.org/u/votti)\
**Replies:** 1\
**Last updated:** [August 16, 2023, 8:49pm UTC](https://discourse.scverse.org/t/how-to-best-integrate-genomic-variant-information/1673 "2023-08-16T20:49:25Z")

</div>

Hi there, I am currently evaluating Muon to provide the backend for multi-omics, patient level (=not single cell) datasets. What I have not found yet is if there is already a recommended way/example how to include geno…

---

## [Debugging Issue for CITE-seq analysis with totalVI juytper notebook tutorial](https://discourse.scverse.org/t/debugging-issue-for-cite-seq-analysis-with-totalvi-juytper-notebook-tutorial/1516)

<div class="topic-metadata">

**Author:** [@Bakar](https://discourse.scverse.org/u/Bakar)\
**Replies:** 6\
**Last updated:** [June 21, 2023, 6:51pm UTC](https://discourse.scverse.org/t/debugging-issue-for-cite-seq-analysis-with-totalvi-juytper-notebook-tutorial/1516 "2023-06-21T18:51:18Z")

</div>

I am having issues addressing a error when trying to go through the CITE-seq analysis with totalVI notebook on my personal computer. An error surfaces when I try to run the following cell block: --------------------…

---

## [COMPASS Installation tutorial - CPLEX Installation](https://discourse.scverse.org/t/compass-installation-tutorial-cplex-installation/1527)

<div class="topic-metadata">

**Author:** [@Bakar](https://discourse.scverse.org/u/Bakar)\
**Replies:** 0\
**Last updated:** [June 19, 2023, 12:55am UTC](https://discourse.scverse.org/t/compass-installation-tutorial-cplex-installation/1527 "2023-06-19T00:55:42Z")

</div>

(Compass is not added to any of the threads/list) Compass homepage: Welcome to Compass’s documentation! — Compass 2021 documentation Compass Github: GitHub - YosefLab/Compass: In-Silico Modeling of Metabolic Heterogene…

---

## [AnnLoader for mudata?](https://discourse.scverse.org/t/annloader-for-mudata/1323)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 1\
**Last updated:** [April 18, 2023, 1:13am UTC](https://discourse.scverse.org/t/annloader-for-mudata/1323 "2023-04-18T01:13:58Z")

</div>

It seems like mudata implements the required methods of the torch.dataset: print(mdata.\_\_len\_\_()) print(mdata.\_\_getitem\_\_(0)) train\_loader = DataLoader( mdata, batch\_size=batch\_size, shuffle=True ) len(trai…

---

## [Question about Github Issue](https://discourse.scverse.org/t/question-about-github-issue/1200)

<div class="topic-metadata">

**Author:** [@maflot](https://discourse.scverse.org/u/maflot)\
**Replies:** 2\
**Last updated:** [March 22, 2023, 3:20pm UTC](https://discourse.scverse.org/t/question-about-github-issue/1200 "2023-03-22T15:20:40Z")

</div>

I have questions about the scanpy foldchange computations. I stumbled across these two issues, which point out two severe issues about the foldchange computation and the tl.rank\_genes\_groups function. Also, I also …

---

## [Make another probabilistic model in scvi-tools](https://discourse.scverse.org/t/make-another-probabilistic-model-in-scvi-tools/902)

<div class="topic-metadata">

**Author:** [@eroell](https://discourse.scverse.org/u/eroell)\
**Replies:** 2\
**Last updated:** [March 22, 2023, 8:36am UTC](https://discourse.scverse.org/t/make-another-probabilistic-model-in-scvi-tools/902 "2023-03-22T08:36:36Z")

</div>

Hi all! Awesome work with this toolbox harmonizing the interfaces and user experience of single cell probabilistic models! We are currently developing such a model, and have interest in implementing it in the scvi-tool…

---

## [Mini-batching error after registering MuData .varm field with custom scvi-tools model](https://discourse.scverse.org/t/mini-batching-error-after-registering-mudata-varm-field-with-custom-scvi-tools-model/1178)

<div class="topic-metadata">

**Author:** [@logan-blaine](https://discourse.scverse.org/u/logan-blaine)\
**Replies:** 1\
**Last updated:** [March 1, 2023, 6:29pm UTC](https://discourse.scverse.org/t/mini-batching-error-after-registering-mudata-varm-field-with-custom-scvi-tools-model/1178 "2023-03-01T18:29:05Z")

</div>

Hi there, I’m trying to build a custom scvi-tools model using the Pyro backend that reads information from a MuData object, including some data from a MuData .varm field. I’ve extended the BaseModelClass with PyroSviTra…

---

## [scVI access the model](https://discourse.scverse.org/t/scvi-access-the-model/1118)

<div class="topic-metadata">

**Author:** [@maflot](https://discourse.scverse.org/u/maflot)\
**Replies:** 4\
**Last updated:** [February 9, 2023, 8:24pm UTC](https://discourse.scverse.org/t/scvi-access-the-model/1118 "2023-02-09T20:24:07Z")

</div>

Hi I was wondering how I would access the different model layers within the scVI class structure. In particular I want to extract the weights from the input layer to the first hidden layer. Has someone an idea how to a…

---

## [How to create a layer over a subset of .obs or .vars?](https://discourse.scverse.org/t/how-to-create-a-layer-over-a-subset-of-obs-or-vars/1070)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 1\
**Last updated:** [January 20, 2023, 1:06pm UTC](https://discourse.scverse.org/t/how-to-create-a-layer-over-a-subset-of-obs-or-vars/1070 "2023-01-20T13:06:32Z")

</div>

I’m using the MultiVI framework for multiomic data which defines modality as a column in adata.var. However, I need to track certain modality-specific transformations in the same multimodal anndata object. But the API …

---

## [Adding "equipment" to AnnData objects for neat addition of functionality](https://discourse.scverse.org/t/adding-equipment-to-anndata-objects-for-neat-addition-of-functionality/588)

<div class="topic-metadata">

**Author:** [@MartinAlvarezKuglen](https://discourse.scverse.org/u/MartinAlvarezKuglen)\
**Replies:** 3\
**Last updated:** [November 1, 2022, 5:48pm UTC](https://discourse.scverse.org/t/adding-equipment-to-anndata-objects-for-neat-addition-of-functionality/588 "2022-11-01T17:48:41Z")

</div>

Hi all, I am adding this topic because my colleague and I had this idea and I was wondering what the community’s suggestion would be for this. In the creation of my ecosystem package I have added quite a few functions …

---

## [How to dynamically mask categorical\_covariate\_keys during training?](https://discourse.scverse.org/t/how-to-dynamically-mask-categorical-covariate-keys-during-training/840)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 3\
**Last updated:** [October 19, 2022, 4:55pm UTC](https://discourse.scverse.org/t/how-to-dynamically-mask-categorical-covariate-keys-during-training/840 "2022-10-19T16:55:14Z")

</div>

I run setup\_anndata(categorical\_covariate\_keys=\['key1','key2',...\]), and need a way to use these covariates selectively in training\_step() (manual optimization). Is there a way to do this? Thanks!

---

## [Progress not reporting loss from custom training plan](https://discourse.scverse.org/t/progress-not-reporting-loss-from-custom-training-plan/765)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 3\
**Last updated:** [October 5, 2022, 11:12pm UTC](https://discourse.scverse.org/t/progress-not-reporting-loss-from-custom-training-plan/765 "2022-10-05T23:12:23Z")

</div>

After training, I am able to view the losses that were computed by running: \>\>\> print(model.trainer.logger.history.items()) dict\_items(\[('train\_loss\_step', train\_loss\_step epoch 0 316.2103…

---

## [Difference between \`TrainingPlan.log()\` vs \`TrainingPlan.compute\_and\_log\_metrics()\`?](https://discourse.scverse.org/t/difference-between-trainingplan-log-vs-trainingplan-compute-and-log-metrics/764)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 1\
**Last updated:** [October 5, 2022, 11:09pm UTC](https://discourse.scverse.org/t/difference-between-trainingplan-log-vs-trainingplan-compute-and-log-metrics/764 "2022-10-05T23:09:36Z")

</div>

What is the difference between TrainingPlan.log() and TrainingPlan.compute\_and\_log\_metrics()? In scvi.train.TraniningPlan, it looks like self.log() just records a subset (the average reconstruction loss) of the stuff re…

---

## [Calls to decoder.forward() in VAE.generative() wrt the current DecoderSCVI API?](https://discourse.scverse.org/t/calls-to-decoder-forward-in-vae-generative-wrt-the-current-decoderscvi-api/755)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 0\
**Last updated:** [September 15, 2022, 6:57pm UTC](https://discourse.scverse.org/t/calls-to-decoder-forward-in-vae-generative-wrt-the-current-decoderscvi-api/755 "2022-09-15T18:57:57Z")

</div>

Looks like VAE.generative() is passing batch\_index, \*categorical\_input, and y to \[already-refactored?\] DecoderSCVI.forward() which is parsing them under a single variable-length parameter \*cat\_list. Was this a design go…

---

## [\`tests/models/test\_pyro.py::test\_pyro\_bayesian\_save\_load\` fails](https://discourse.scverse.org/t/tests-models-test-pyro-py-test-pyro-bayesian-save-load-fails/713)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 2\
**Last updated:** [August 26, 2022, 5:56pm UTC](https://discourse.scverse.org/t/tests-models-test-pyro-py-test-pyro-bayesian-save-load-fails/713 "2022-08-26T17:56:13Z")

</div>

As a sanity check, I ran cd scvi-tools && pytest within my new build environment and got the following: ===================================================================================================================…

---

## [Help understanding inheritance for scvi.models](https://discourse.scverse.org/t/help-understanding-inheritance-for-scvi-models/699)

<div class="topic-metadata">

**Author:** [@mkarikom](https://discourse.scverse.org/u/mkarikom)\
**Replies:** 1\
**Last updated:** [August 22, 2022, 9:13pm UTC](https://discourse.scverse.org/t/help-understanding-inheritance-for-scvi-models/699 "2022-08-22T21:13:36Z")

</div>

I’m looking into implementing a new project as an scvi model and am working my way through the code base to understand a bit more what is going on. I’m still new to python module development and I’m wondering why certai…

[Next page](https://discourse.scverse.org/tag/developer/24.md?match_all_tags=true&page=1&tags%5B%5D=developer)