WormBase anndata standard wrangling conventions

This document describes how WormBase should wrangle single cell RNA seq data into the anndata format in .h5ad files with standard fields, plus any number of optional fields that vary depending on the metadata the authors provide.

As possible, we attempt to keep the field names lower case, short, descriptive, and only using valid Python variable names so they may be accessed via the syntax adata.var.field_name

Our goal is to standartize the naming convention for frequently used fields so that code and pipelines may be reused without headaches changing variable names. The .h5ad file should only contain genes and cells with at least one count.

We are trying to keep these guidelines simple so that we can easily stick to them. Going forward we will update them as we learn from experience. For example, at some point WormBase may start uniformly reprocessing the data, so a version field may be needed in the adata.uns

Below we provide a standard description of the mandatory fields we use in all datasets, plus some common optional ones that we have used so far (not all, each dataset may have different metadata provided by the authors).

adata.var: gene IDs, names and descriptions

Field name	Description	Type	Example value	Optionality
adata.var.index	WormBase gene ID, must be unique	string	WBGene00010957	Required
adata.var.gene_id	WormBase gene ID, repeat values from index	string	WBGene00010957	Required
adata.var.gene_name	WormBase gene name	string	nduo-6	Required
adata.var.gene_description	WormBase short gene description. Full list available for download here	string	Predicted to have NADH dehydrogenase (ubiquinone) activity. Predicted to localize to integral component of membrane; mitochondrial membrane; and respirasome.	Optional

adata.obs: cell barcode, experiment, batch, original study, cell type

Field name	Description	Type	Example value	Optionality
adata.var.index	The batch name joined with cell barcode witha + char	string	F4_1+TGTAACGGTTAGCTAC-1	Required
adata.var.study	A unique shorthand for the study that published the data, ideally in the style all lower case. The .h5ad file should have the same name as the study it corresponds to.	categorical	taylor2020	Required
adata.obs.sample_batch	The run that produced the corresponding barcode. Usually sample_batch and sample will be the same, but with multiplexing one batch can have multiple samples	categorical	F4_1	Required
adata.obs.sample	The name of the biological sample that is in this batch	categorical string	L2 larvae fourth repeat	Required
adata.obs.sample_description	Description of the sample. This is mandatory because otherwise it will be very easy to confuse two samples from their name without carefully reading the paper or contacting authors	categorical string	F4_1	Required
adata.obs.barcode	The cell barcode	string	AAACCCAAGATCGCTT-1	Required
adata.obs.cell_type	The cell type annotation provided by the authors. Should be unlabeled if not available	categorical	Neuronal	Required
adata.obs.cell_subtype	The cell subtype annotation if provided by the authors. If the annotation contains individual neuron types, these should be cell subtypes. If only a cell type label was provided, it can be repeated in both cell_type and cell_subtype	categorical	ASJ	Required