Curate neurotransmitter type for human brain neuron types in Siletti

[dosumis]

DRAFT:

Curate neurotransmitter type for human brain neuron types in Siletti et al. (2022) bioRxiv https://www.biorxiv.org/content/10.1101/2022.10.12.511898v1

TODO: @dosumis @hkir-dev to document how to record (what neuron type/cluster IDs to link with)

[hkir-dev]

In the whole human brain ontology (whbo) project I added a Neurotransmitter column to the curation table: https://github.com/hkir-dev/whole_human_brain_ontology/blob/main/src/patterns/data/default/CS202210140_class_curation.tsv

These annotations are rolled using the following dosdp pattern:

   - axiom_type: subClassOf
     multi_clause:
       sep: " and "
       clauses:
         - text: "'capable_of' some %s"
           vars:
             - Neurotransmitter

If this is appropriate, we can use this curation table and hopefully whbo will be part of the PCL soon. Or I can add a similar logic to the brain data standards curation tables as well.

[dosumis]

The task here is to map cell types from the paper to neurotransmitters.

Glutamatergic vs GABAergic can be curated from figure 2

To record NT - add a term from the GO 'neurotransmitter secretion' branch to the Neurotransmitter column of the TSV.

(the design pattern above adds a capable_of axiom using this which => autoclassification using the reasoner.

Mapping: The prefLabel column has entries like this: DLIT_142. The number maps to the cluster number in the figure. (Deep IT) 142 => DLIT_142.

There may be other figures or tables or text that we could use to map other NTs. Please review text for these.

[hkir-dev]

By the way Neurotransmitter column supports multi-value. | separated multiple GO terms can be set as value (such as GO:0061534|GO:0061527).

[dosumis]

Update: NT mappings are in Table S2 downloadable from supplementary-material.

All we need is to map the abbreviations used in this table to terms in the GO neurotransmitter secretion branch. It would also be useful to review the paper for information about what markers were used to infer cell type. I suspect that VGLUT2 vs VGLUT3 etc may refer to different markers (likely transporters).

Output: A table mapping abbreviation to GO terms to evidence comment of the form: "inferred to be {x}-ergic based on expression of {y}.

[ghost]

It would also be useful to review the paper for information about what markers were used to infer cell type.

On cursory review, it appears the genes noted may be specific to cell types or clusters and not necessarily used to identify all neurotransmitter-specific cells. For example:

"The splatter supercluster uniquely comprised both inhibitory and excitatory cells: 22.1% of cells expressed the GABA transporter SLC32A1, 39.3% of cells expressed the predominant glutamatergic transporter SLC17A6, and 1.4% of cells co-expressed these two genes (Fig. 3B; Fig. S4C)."

Would it be appropriate to assign SLC32A1 to all GABAergic neurones?

SYMBOL	CELL TYPE LABEL	CELL TYPE NEUROTRANSMISSION ID	GENES
GABA	GABAergic	GO:0061534	SLC32A1

[scheuerm]

My understanding is that neurotransmitters are proteins that are secreted from one cell to transmit a signal to another cell across a synapse. SLC32A1 is a transporter that takes up certain neurotransmitters into synaptic vesicles. So, it is involved in the neurotransmission process but is not a neurotransmitter. We should be careful about the distinction.

[ghost]

Thank you, @scheuerm.

Neurotransmitters (NTs) are molecules, not necessarily proteins, that can function as you describe. The idea is not to suggest that the gene is the NT, but rather the gene expression was used to identify what NT a neurone would release. For example, expression of SLC17A8 may be used to identify a glutamatergic neuron since this gene encodes a transporter protein used to package glutamate into synaptic vesicles. When an action potential occurs, the neurone would then release glutamate from the vesicles into the synaptic cleft. Since the neurone releases glutamate as a NT, it would be classified as glutamatergic.

I still caution against the broad application of the pattern "inferred to be {x}-ergic based on expression of {y}" because it may only apply to certain clusters and not all {x}-ergic neurones.

[dosumis]

It would also be useful to review the paper for information about what markers were used to infer cell type.

On cursory review, it appears the genes noted may be specific to cell types or clusters and not necessarily used to identify all neurotransmitter-specific cells. For example:

"The splatter supercluster uniquely comprised both inhibitory and excitatory cells: 22.1% of cells expressed the GABA transporter SLC32A1, 39.3% of cells expressed the predominant glutamatergic transporter SLC17A6, and 1.4% of cells co-expressed these two genes (Fig. 3B; Fig. S4C)."

Would it be appropriate to assign SLC32A1 to all GABAergic neurones?

SYMBOL	CELL TYPE LABEL	CELL TYPE NEUROTRANSMISSION ID	GENES
GABA	GABAergic	GO:0061534	SLC32A1

Good point. Are there any clues in the methods section?

From this bit of text, it looks like we know the marker for 'splatter' neurons so it would be safe to record as evidence for these neurons. There may be other cases where the text is clear about specific clusters. We could keep evidence recording to these cases until we have more info. I would also like to collect a set of questions that we can ask the authors. I think we should ask whether multiple markers were used for NT calling and if so, whether they can provide a mapping to clusters.

[hkir-dev]

Neurotransmitter curation applied to the whole human brain ontology except the evidence comments since we don't have genes yet. _New location of the NT mapping table is: https://github.com/hkir-dev/whole_human_brain_ontology/blob/main/src/dendrograms/supplementary/Neurotransmitter_symbols_mapping.tsv_