Biology Atlas

Map / Concepts

concept · intermediate · partly written

Assay metadata as part of the representation: batch labels and what has been shown to help

Methods that use assay metadata use one field, a batch label, and usually subtract it; metadata such as operator, site, date or quality as typed content of the representation has no measured gain on record.

His idea, in his words

"It would be cool to invent new ontologies that somehow take advantage of different assay metadata, because assays are done by different people in different places at different times and have different longitudinality and data quality." 1

Amit's hypothesis

The idea stated so it can be wrong more

Operator, site, date, longitudinality and quality, as typed fields used as model inputs or edge weights, beat a batch label alone. Our reading of his words. 23

Our inference

What exists more

MIAME argued in 2001 that reported microarray data were stripped of the evidence needed to judge their quality. The working fix uses one categorical variable. ComBat applies an empirical-Bayes adjustment that is robust in small batches; Harmony and scVI take the batch label as input. Luecken and colleagues tested 68 method-and-preprocessing combinations on 85 batches from 23 publications; the metrics score batch mixing against cell-type labels that annotators made. The label is the only link to provenance, and the methods delete the effect it names. 4352

Reported, not yet replicated

Where it fails more

When batch is tangled with the group of interest, correction inflates confidence: in a reanalysis of an unbalanced dataset, ComBat with the treatment as covariate gave 2,011 genes at 5% false discovery rate against 11 when batch was simply blocked. Single-cell integration can erase real biology, and the benchmark metrics reward mixing. Separately, assay details are mostly not reported in public bioactivity databases, and IC50 spread was only about 25% larger than Ki spread, a moderate amount of noise. 627

Reported, not yet replicated

What remains open (our inference) more

Operator, site, date or longitudinality as typed attributes with a measured gain over the batch label alone: not found. Quality or longitudinality as an edge attribute: not found. Provenance vocabularies such as PROV-O record who and what produced data; nothing opened shows them improving a prediction. 27

Our inference

Cheapest test (our inference) more

On a public multi-lab ADMET or CRISPR set where batch is balanced against condition, add source lab, assay type and year as features and compare held-out-lab prediction with no correction and with the batch label alone, splitting by lab; a random split would flatter it. If the richer fields add nothing over the batch label, the ontology idea has shown nothing. Fail fast if the fields are missing. 67

Our inference

Still open here

Not yet written: Plain answer, Why it matters, Equation, Real numbers.

Linked from

Sources

  1. Amit's dictation of 2026-11-18: how biology data is represented, and where ontologies can be innovated. Kept in the atlas repository; not published.
  2. Luecken MD, et al.. Benchmarking atlas-level data integration in single-cell genomics. Nature Methods (2022). paper; opened 2026-11-18 all records citing this
  3. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics (2007). paper; opened 2026-11-18 all records citing this
  4. Brazma A, Hingamp P, Quackenbush J, et al.. Minimum information about a microarray experiment (MIAME)—toward standards for microarray data. Nature Genetics (2001). paper; opened 2026-11-18 all records citing this
  5. Korsunsky I, Millard N, et al.. Fast, sensitive and accurate integration of single-cell data with Harmony. Nature Methods (2019). paper; opened 2026-11-18 all records citing this
  6. Nygaard V, Rødland EA, Hovig E. Methods that remove batch effects while retaining group differences may lead to exaggerated confidence in downstream analyses. Biostatistics (2016). paper; opened 2026-11-18 all records citing this
  7. Kalliokoski, T., et al.. Kalliokoski et al. 2013: Comparability of mixed IC50 data (PLoS ONE). PLoS ONE (2013). paper; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (5)

This record has been changed 5 times, 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
2026-11-19
Record
content/concepts/assay-metadata-in-the-representation.json