=Paper=
{{Paper
|id=Vol-2807/abstractV
|storemode=property
|title=Current OBO Ontologies Are Not Sufficient to Annotate COVID-19-Related Cell Types (short paper)
|pdfUrl=https://ceur-ws.org/Vol-2807/abstractV.pdf
|volume=Vol-2807
|authors=Tiago Lubiana
|dblpUrl=https://dblp.org/rec/conf/icbo/Lubiana20a
}}
==Current OBO Ontologies Are Not Sufficient to Annotate COVID-19-Related Cell Types (short paper)==
https://ceur-ws.org/Vol-2807/abstractV.pdf
Current OBO ontologies are not sufficient
to annotate COVID-19-related cell types
Tiago LUBIANA a,1
a University of São Paulo, Brazil
Abstract. The OBO Foundry ontologies are outstanding resources for classifying
and curating concepts in the life sciences. However, the use of annotated texts, fig-
ures, and data is far from widespread. In this work, I curated 360 cell type men-
tions in five different single-cell RNA-seq, COVID-19 related scientific articles.
These mentions were gathered from the main figures, alongside mentions of tissue,
taxon, life stage, and gender of the samples. I manually matched the terms used to
NCBITaxon, UBERON, MMUSDV, HSAPDV, PATO, UBERON, and CL ontolo-
gies where appropriate. Only 130/360 cell type mentions (36%) could be matched
(based on synonyms) to UBERON and CL. Strikingly, none of the 360 mentions
cell types could be completely described by OBO ontologies. These results provide
evidence that there is an urgent need to improve the reporting of cell-type-related
results and enhance ontology generation systems to facilitate completeness in the
light of the rise of novel definitions.
Keywords. Cell Ontology, cell types, COVID-19, single-cell RNA-seq
The Cell Ontology (CL) is a long ongoing effort and an outstanding resource for
the scientific community. It focuses on general cell types (e.g. ”neuron”, without species
restriction) and it is interoperable with the other OBO Foundry ontologies for taxons, tis-
sues, stages of life, and more. It is expected, then, that a combination of these ontologies
should be enough to annotate cell types described in scientific articles. The COVID-19
pandemic has appeared in the moment of the rising use of single-cell RNA sequencing.
Cell types are at the core of single-cell transcriptomics. Articles that sequence individual
cells in the contexto of COVID-19 are rapidly being published, and annotation of data
and results is important to ensure interoperability and data reuse. This work explores the
challenges of annotating these articles using current OBO ontologies.
I selected five articles about COVID-19 that use single-cell transcriptomics as the
main experimental approach to investigate this. Two of them [1][2] studied airway cells,
the other two investigated peripheral blood immune cells [3][4] and a fifth investigated
cells in 13 different human tissues[5]. Cell types were identified on main text figures,
alongside four core descriptors (tissue of origin, taxon, age, and biological sex of sam-
ples). Data was curated in a spreadsheet and were manually searched in Ontobee in
the NCBITaxon, UBERON, MMUSDV, HSAPDV, PATO, UBERON, and CL ontologies
where appropriate.
I identified 360 mentions to cell types spread across 19 figures in the 5 papers. From
these 360 mentions, I was able to find matches on CL for only 238 (66%). In other words,
1 Corresponding Author: Tiago Lubiana, University of São Paulo, Brazil. E-mail:tiago.lubiana.alves@usp.br
Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
roughly a third of cell types in the dataset are not represented in CL, if we reconcile both
sources by using synonyms.
The situation is worsened because the definitions of cell types on the Cell Ontology
do not allow precise classification based solely on transcriptomic characteristics. Rigor-
ously, none of the cell types claimed in these studies are properly captured by CL. For
example, the definition of a dendritic cell (CL 0000451) starts by “A cell of hematopoi-
etic origin, typically resident in particular tissues (...)”. All five articles use “dendritic
cell” (or one of its subclasses) to label single-cell clusters, and none of those clusters can
be rigorously matched to CL, as the articles do not trace their origin.
We know that cell types behave differently depending on tissue, biological sex, and
life stage. I could label both the tissue and cell line for 130/360 cell type mentions
(36%). Only 35/360 (9%) had statements that could be matched to life stage ontologies
(MMUSDV, HSAPDV). None of the 360 cell types mentions had annotations regarding
the biological sex of the subjects. This lack of consideration of biological sex, and little
consideration of age is striking. On the bright side, all 360 mentions included the taxon
of interest, allowing precise matching to NCBITaxon ontology.
In light of the Human Cell Atlas [6] and the considerable effort to characterize cell
types, it becomes urgent to advance the use of ontologies to annotate these texts. Due to
a lack of accuracy in reporting, the gaps in the Cell Ontology (and other ontologies), and
the challenges of defining cell types, it is not possible to accurately annotate single-cell
transcriptomic studies. The intent of this work is to bring these issues into attention, as
annotation would be vital for maximizing what we can get from the data.
Systems such as Wikidata might complement the OBO Foundry effort by allowing
public edits and avoiding formal definitions. By using descriptions instead of definitions,
we can work with cell types that are rigorously different, but in practice treated as the
same. The putative lack of rigor of such systems might be outweighed by the benefits of
making annotations feasible. Also, by providing user-friendly graphic interfaces, Wiki-
data might make it easier for more biologists to get initiated in ontologies, possibly im-
proving, in the long run, the precision of the terms used in scientific reporting.
References
[1] Muus C, Luecken MD, Eraslan G, Waghray A, Heimberg G, Sikkema L, Kobayashi Y, Vaishnav ED,
Subramanian A, Smilie C, Jagadeesh K. Integrated analyses of single-cell atlases reveal age, gender, and
smoking status associations with cell type-specific expression of mediators of SARS-CoV-2 viral entry
and highlights inflammatory programs in putative target cells. BioRxiv. 2020 Jan 1.
[2] Liao M, Liu Y, Yuan J, Wen Y, Xu G, Zhao J, Cheng L, Li J, Wang X, Wang F, Liu L. Single-cell
landscape of bronchoalveolar immune cells in patients with COVID-19. Nature Medicine. 2020 May
12:1-3.
[3] Wen W, Su W, Tang H, Le W, Zhang X, Zheng Y, Liu X, Xie L, Li J, Ye J, Dong L. Immune cell
profiling of COVID-19 patients in the recovery stage by single-cell sequencing. Cell Discovery. 2020
May 4;6(1):1-8.
[4] Wilk AJ, Rustagi A, Zhao NQ, Roque J, Martinez-Colon GJ, McKechnie JL, Ivison GT, Ranganath
T, Vergara R, Hollis T, Simpson LJ. A single-cell atlas of the peripheral immune response to severe
COVID-19. medRxiv. 2020 Jan 1.
[5] Qi F, Qian S, Zhang S, Zhang Z. Single cell RNA sequencing of 13 human tissues identify cell types and
receptors of human coronaviruses. Biochemical and biophysical research communications. 2020 Mar
19.