chemical-plant sparql query

[YojanaGadiya]

Dear EMIKG team,

Since I have limited SPARQL query formulation skills, I was wondering if you or the community could assist me in building one to extract chemical-plant relations from the graph.

Additionally, on looking into the underlying schema of the graph, it looked like these edges come from WikiData. Is that correct?

Thank You.

Regards, Yojana

[oolonek]

Hi @YojanaGadiya,

Thanks for getting in touch.

The Examples section of the Wikidata Query Service https://query.wikidata.org/ is an excellent ressource to start learning SPARQL.

More specifically if you want to query the ENPKG a good starting point would be for you to explore the queries presented in Table 1 of the preprint (https://doi.org/10.26434/chemrxiv-2023-sljbt-v2). Do let us know if you need help in adapting some of them. Some nodes correspond indeed to Wikidata objects. E.g. for a taxon https://enpkg.commons-lab.org/graphdb/graphs-visualizations?saved=763f2599add54363b4fdcfe4830ac63a or a molecule https://enpkg.commons-lab.org/graphdb/graphs-visualizations?saved=ac772b1f386a42b69b7786d632c12a7a.

[YojanaGadiya]

Hi @oolonek,

These are great pointers! Just one last question. Were there any stats on the difference between the EMIKG and LOTUS specifically on the taxon-molecule edges?

Regards, Yojana Gadiya

[oolonek]

Hi @YojanaGadiya ,

At the moment the EMIKG is under construction. I define here these acronyms as I understand they can be misleading.

• ENPKG (Experimental Natural Products Knowledge Graph) is the name of a KG that has been build as a first pilot in order to explore the conversion of untargeted metabolomics datasets acquired from plant extracts in Linked Open Data. This KG currently contains data acquired from :

  • 1600 tropical plant extracts (a datset previously described at https://doi.org/10.1093/gigascience/giac124)
  • 3 additional Waltheria indica extract (a plant profiled in our lab under different conditions and with different extraction methods)
  • 337 Korean plant extracts (a dataset acquired by Kyo Bin Kang's team, described at https://doi.org/10.1038/s41597-022-01662-2 and publicly available at https://doi.org/doi:10.25345/C5SB50. So profiled under different conditions and with different extraction methods, by a different team. This was to showcase the potential to align datasets from different sources.

• EMIKG (Earth Metabolome Initiative Knowledge Graph) is the name of the KG we are currently constructing in the frame of the Earth Metabolome Initiative. Currently we are building the data transformation portal (https://github.com/earth-metabolome-initiative/emi-monorepo) which will allow users to convert metabolomics datasets and the associated metadata in a graph. At the moment this KG is not populated.

• LOTUS is the name of an Open electronic ressource describing natural products occurences. It is fully described here (https://doi.org/10.7554/eLife.70780) and available through the Wikidata graph. Do not rely on the https://lotus.naturalproducts.net/ website as it is not maintained anymore. The https://search.nprod.net/ portal is beeing build at the moment by @bjonnh and @adafede and will gradually replace the https://lotus.naturalproducts.net/ website. In any cases, Wikidata is the reference source for LOTUS data.

I hope this helps. Also to keep in mind, the objective of ENPKG/EMIKG is to gather and organize massive amount of putative biological source - chemical structures links established through mass spectrometry based metabolomics experiments. The purpose of LOTUS is to gather biological source - chemical structures links supported by a scientific publication and usually obtained through a physical isolation of the molecule from the biological source.

Regarding current stats here are some figures:

• LOTUS : Table 3, Figure 4 and 5 should give and overview of the situation at the time of publication. SPARQL queries at Wikidata should be used to retrieve the current stats. E.g. https://w.wiki/8wqq returns 650,955 distinct chemical structure / taxon / refs as of today.

• ENPKG : You can run similar queries on the graph. E.g. Return distinct wd_taxon wd_chemical pairs annotated through Sirius Currently this returns 901,352 distinct wd_taxon wd_chemical pairs annotated through Sirius. Note that these is overevaluated since here we link all possible 3D structures to the 2D structures (the latter beeing the best we can reach with mass spec annotations). If you run the query for distinct 2D structures (Return distinct wd_taxon IK2D pairs annotated through Sirius) you will get 368,327 distinct wd_taxon wd_chemical pairs annotated through Sirius.

I hope this helps. Let us know if you have any further question.