Add group-wise miQC modeling for multimodal mitochondrial fractions

[Copilot]

miQC previously fit one global 2-component mixture, which can misclassify cells when datasets contain biologically distinct cell types with different baseline %mtRNA. This update introduces optional group-wise modeling/filtering so posterior-based QC is applied within relevant subpopulations.

• Modeling: optional per-group fits

  • Added group to mixtureModel().
  • When group is set to a colData(sce) column, miQC fits one model per group and returns a named grouped model object.
  • Default behavior is unchanged (group = NULL uses a single global model).

• Filtering: optional per-group posterior decisions

  • Added group to filterCells().
  • Supports grouped models from mixtureModel(..., group=...) and applies posterior cutoff + boundary rules independently per group before recombining kept cells.
  • Added validation for missing/invalid group columns and mismatched model/group usage.

• Documentation updates

  • Updated mixtureModel and filterCells Rd docs for new group parameter and grouped return behavior.
  • Added vignette section showing recommended usage for mixed cell populations with different baseline mitochondrial fractions.
  • Added NEWS entry describing group-wise support.

# Apply miQC within annotated cell-type strata
model_by_type <- mixtureModel(sce, group = "cell_type")
sce <- filterCells(sce, model = model_by_type, group = "cell_type")

[cgreene]

This should really be a new version, not an additional bullet point in 1.7.1. Can you position it this way?

[Rfriedmancolumbia]

Casey,
Thank you.
Please excuse my ignorance of github.
How do I get the script that does the multimodal analysis?

Thanks and best wishes,
Rich

[cgreene]

Hi Rich,

If you are not a heavy github user this might be tough. What you are looking to do is clone the pull request's branch locally. The information on the branch is above (copilot/multimodal-mtrna-filter-cells) but you probably would need to be reasonably experienced with github to navigate this in practice (or an aggressive googler ;)).

Thanks!
Casey

[Rfriedmancolumbia]

Casey, Thanks. I will give it a shot, maybe not today. Best wishes, Rich

________________________________ From: Casey Greene ***@***.***> Sent: Wednesday, May 27, 2026 12:54 PM To: greenelab/miQC ***@***.***> Cc: Friedman, Richard A. ***@***.***>; Comment ***@***.***> Subject: [EXTERNAL] Re: [greenelab/miQC] Add group-wise miQC modeling for multimodal mitochondrial fractions (PR #18) @cgreene commented on this pull request.

________________________________ In NEWS<#18 (comment)>:

@@ -12,3 +12,4 @@ Changes in version 1.1.5 (2021-08-24)

+ Updated citation Changes in version 1.7.1 (2023-01-04) + Added new function, get1DCutoff ++ Added optional group-wise modeling/filtering support to handle mixed cell populations with different baseline mitochondrial fractions Hi Rich, If you are not a heavy github user this might be tough. What you are looking to do is clone the pull request's branch locally. The information on the branch is above (copilot/multimodal-mtrna-filter-cells) but you probably would need to be reasonably experienced with github to navigate this in practice (or an aggressive googler ;)). Thanks! Casey — Reply to this email directly, view it on GitHub<#18?email_source=notifications&email_token=AHT5YEROO2EJR4G7I2NWMDT444MVLA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTIMZXGQZTIMZZHA3KM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#discussion_r3312426501>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AHT5YEW7H6NNWOE5DMLZX5L444MVLAVCNFSM6AAAAACZPKN62CVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHM2DGNZUGM2DGOJYGY>. Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS<https://github.com/notifications/mobile/ios/AHT5YERQEEEDNFIT55OJCGD444MVLA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTIMZXGQZTIMZZHA3KM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJKTGN5XXIZLSL5UW64Y> and Android<https://github.com/notifications/mobile/android/AHT5YEQ7ZZJN7LGDRSIC6OT444MVLA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTIMZXGQZTIMZZHA3KM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>. Download it today! You are receiving this because you commented.Message ID: ***@***.***>

[Rfriedmancolumbia]

Casey, Upon mature reflection, I don’t think that I should use the multimodal version unless and until it is peer reviewed. I am a bioinformatics consultant for a Core at Columbia, and I have to justify everything I do in terms of established methods. I thank both you and Co-Pilot for your efforts. Best wishes, Rich From: Casey Greene ***@***.***> Date: Wednesday, May 27, 2026 at 12:54 PM To: greenelab/miQC ***@***.***> Cc: Friedman, Richard A. ***@***.***>; Comment ***@***.***> Subject: [EXTERNAL] Re: [greenelab/miQC] Add group-wise miQC modeling for multimodal mitochondrial fractions (PR #18) @cgreene commented on this pull request.

________________________________ In NEWS<#18 (comment)>:

@@ -12,3 +12,4 @@ Changes in version 1.1.5 (2021-08-24)

+ Updated citation Changes in version 1.7.1 (2023-01-04) + Added new function, get1DCutoff ++ Added optional group-wise modeling/filtering support to handle mixed cell populations with different baseline mitochondrial fractions Hi Rich, If you are not a heavy github user this might be tough. What you are looking to do is clone the pull request's branch locally. The information on the branch is above (copilot/multimodal-mtrna-filter-cells) but you probably would need to be reasonably experienced with github to navigate this in practice (or an aggressive googler ;)). Thanks! Casey — Reply to this email directly, view it on GitHub<#18?email_source=notifications&email_token=AHT5YEROO2EJR4G7I2NWMDT444MVLA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTIMZXGQZTIMZZHA3KM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#discussion_r3312426501>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AHT5YEW7H6NNWOE5DMLZX5L444MVLAVCNFSM6AAAAACZPKN62CVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHM2DGNZUGM2DGOJYGY>. Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS<https://github.com/notifications/mobile/ios/AHT5YERQEEEDNFIT55OJCGD444MVLA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTIMZXGQZTIMZZHA3KM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJKTGN5XXIZLSL5UW64Y> and Android<https://github.com/notifications/mobile/android/AHT5YEQ7ZZJN7LGDRSIC6OT444MVLA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTIMZXGQZTIMZZHA3KM4TFMFZW63VHMNXW23LFNZ2KKZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>. Download it today! You are receiving this because you commented.