concept · intermediate · partly written
What a missing value means in each omics layer
The same blank is a measured zero, an unmeasured feature, a missing layer or an imputed value depending on the layer and tool, and the format rarely says which.
Summary
In AnnData an outer join writes a never-measured feature as 0; in MuData a cell absent from a modality is a 0 in the map; MOFA+ treats a missing layer as unobserved in the likelihood; totalVI zero-fills a missing protein and ignores it in the loss; in sparse ATAC a zero mostly means not detected. The same-looking blank carries several meanings. 12345
Reported, not yet replicated
Tables and containers more
AnnData encodes missing as -1 for categoricals, a boolean mask for nullable types and NaN in floats, and its specification says nothing on NaN in the numeric matrix X. The scverse concatenation tutorial says features missing from one object are filled with 0 instead of na. The MuData specification says an observation or variable missing from a modality corresponds to a 0. 612
Reported, not yet replicated
Methods more
MOFA+ accounts for missing values in its likelihood. totalVI zero-fills a missing protein at input and ignores it in the loss. For sparse binarised ATAC in the NeurIPS challenge, predicting mostly zeros was effective for RMSE, so a do-nothing answer scored well. 345
Reported, not yet replicated
Is a zero in droplet RNA a dropout? more
Negative-control droplets fit a negative binomial without zero inflation in four of five datasets, a preprint by Svensson, while a Bayesian model-selection analysis by Gong and colleagues finds zero inflation mainly biological. The two do not agree and the second was read as a summary. 78
Contested
Imputation creates signal more
Andrews and Hemberg report that the imputation methods MAGIC and dca produced false gene-gene correlations above r = 0.75 in simulation. 9
Reported, not yet replicated
Metabolites more
In mzTab-M a metabolite identifier can denote several candidates by design, so the ambiguity is in identity rather than in the value. 10
Reported, not yet replicated
Plain answer
Why it matters
A practitioner reads a missing value as a measurement; a join that treats it as zero can create signal or hide it. 9
Our inference
Real numbers
False correlations from imputation 2026-11-18
Above r = 0.75 between genes in simulation after MAGIC or dca. 9
Reported, not yet replicated
Still open here
- What a missing value means in proteomics intensity tables and in spatial tables was not opened.
- Standing question Q3 and Q4 as defined in the atlas concern protein structures and compounds; this record answers only the report's own wording.
Not yet written: Equation.
Linked from
Sources
- scverse tutorial: concatenating AnnData and MuData objects.. web, not peer reviewed; opened 2026-11-18 all records citing this
- MuData specification [RFC].. web, not peer reviewed; opened 2026-11-18 all records citing this
- Argelaguet R, Arnol D, Bredikhin D, Deloro Y, et al.. MOFA+: a statistical framework for comprehensive integration of multi-modal single-cell data. Genome Biol (2020). paper; opened 2026-11-18 all records citing this
- Gayoso A, Steier Z, Lopez R, Regier J, et al.. Joint probabilistic modeling of single-cell multi-omic data with totalVI. Nat Methods (2021). paper; opened 2026-11-18 all records citing this
- Multimodal single cell data integration challenge: Results and lessons learned. Proceedings of Machine Learning Research 176 (2022). paper; opened 2026-11-18 all records citing this
- On-disk format (anndata documentation).. web, not peer reviewed; opened 2026-11-18 all records citing this
- Droplet scRNA-seq is not zero-inflated. bioRxiv (2019). preprint, not peer reviewed; opened 2026-11-18 all records citing this
- Choi K, Chen Y, Skelly DA, Churchill GA. Bayesian model selection reveals biological origins of zero inflation in single-cell transcriptomics. Genome Biol (2020). paper; opened 2026-11-18 all records citing this
- Andrews TS, Hemberg M. False signals induced by single-cell imputation. F1000Res (2018). paper; opened 2026-11-18 all records citing this
- Hoffmann N, Rein J, Sachsenberg T, Hartler J, et al.. mzTab-M: A Data Standard for Sharing Quantitative Results in Mass Spectrometry Metabolomics. Anal Chem (2019). paper; opened 2026-11-18 all records citing this
Where it sits on the map
History of this record (1)
This record has been changed 1 time, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.
- Status
- partly written
- Last updated
- 2026-11-18
- Record
- content/concepts/what-a-missing-value-means.json