FAIR DATA Fund use case: evolution of specialised metabolite decorations in brassicaceae

Authors: Feli­cia Wolters, Tina Woldu, Sibbe Bakker, Justin J. J. van der Hooft, Klaas Bouwmeester, Marnix H. Mede­ma; Wagenin­gen Uni­ver­si­ty & Research.

Gen­er­al back­ground

Plants have evolved a pletho­ra of spe­cialised metabo­lites pro­vid­ing valu­able resources for human med­i­cine, nutri­tion and agro­chem­istry. Bioac­tiv­i­ty of spe­cial­ized metabo­lites is ulti­mate­ly depen­dent on high­ly vari­able mod­i­fi­ca­tions of core struc­tures. In par­tic­u­lar, com­pound gly­co­sy­la­tion deter­mines tox­i­c­i­ty con­ferred by the addi­tion of sug­ar moi­eties on vari­able sites of the core struc­ture. Genes encod­ing for UDP-depen­dent gly­co­syl­trans­feras­es (UGTs) evolve rapid­ly via tan­dem dupli­ca­tion and translo­ca­tion, result­ing in the for­ma­tion of UGT gene tan­dem-arrays. Although gly­co­sy­la­tion pro­files con­vey key fea­tures of plant spe­cial­ized com­pound bioac­tiv­i­ty, func­tion­al anno­ta­tion of UGT genes and tra­jec­to­ries of tan­dem-array evo­lu­tion are still large­ly obscure. To this end, inte­gra­tion of genomics, tran­scrip­tomics, and metabolomics data is need­ed. How­ev­er, paired omics datasets are unavail­able for most plant species. 

Approach

In our project, we gen­er­at­ed paired tran­scrip­tomics and metabolomics datasets for close and dis­tant rel­a­tives of the Bras­si­caceae fam­i­ly with pub­licly avail­able genome assem­blies.

Untar­get­ed metabolomics data and paired RNAseq data has been gen­er­at­ed for a bio pan­el of 17 species, con­sist­ing of close and dis­tant rel­a­tives in the Bras­si­caceae fam­i­ly. Based on the eval­u­a­tion of meta­bol­ic dis­tance, a sub­set of ten species was select­ed for the gen­er­a­tion of paired tran­scrip­tomics and metabolomics data in a time-series, includ­ing plant hor­mone treat­ments for elic­i­ta­tion of spe­cial­ized biosyn­thet­ic path­ways. 

Key results

A promi­nent inte­gra­tion method for dif­fer­ent data types has been intro­duced recent­ly cen­tered around metabolomics data analy­sis [1]. The qual­i­ty and com­pre­hen­sive­ness of sam­ple meta­da­ta deter­mines the extent to which the relat­ed­ness of extract­ed fea­tures can be com­pu­ta­tion­al­ly inferred. Har­mo­niza­tion of meta­da­ta stan­dards has been pro­posed recent­ly for the large pub­lic metabolomics repos­i­to­ry Mas­sIVE to enable inte­gra­tion with the Sequence Read Archive (SRA) [2]. 

1. Refine­ment of meta­da­ta struc­ture
   Based on the pro­posed struc­tur­ing of meta­da­ta [1, 2]  and the use of com­mon ontolo­gies for meta­da­ta fields [1 ‑3], we have con­struct­ed a meta­da­ta frame­work for both gen­er­at­ed paired omics datasets with­in this project.Tthe design of this frame­work aims to facil­i­tate fur­ther inte­gra­tion of datasets and paired data types, such as genome anno­ta­tion data, pro­teomics data, and phe­nomics data by com­mon ontol­ogy terms accord­ing to the Planteome knowl­edge­base [3]. 
2. Con­struc­tion of a knowl­edge-graph frame­work for data inte­gra­tion
   Based on the refine­ment of meta­da­ta struc­ture, we have con­struct­ed a pro­to­type for a knowl­edge-graph accord­ing to the Exper­i­men­tal Plant Nat­ur­al Prod­ucts Knowl­edge-graph frame­work intro­duced by Gaudry et al [4]. The meta­da­ta struc­ture was con­vert­ed into a share­able trip­ple for­mat enabling the use of SPARQL queries for extrac­tion of data and relat­ed meta­da­ta.
   

Fig. 1: Sim­pli­fied scheme of meta­da­ta inte­gra­tion for cross-link­ing pub­lic knowl­edge data­bas­es and con­struc­tion of an Exper­i­men­tal Nat­ur­al Prod­ucts Knowl­edge-graph fol­low­ing the con­cept intro­duces by Gaudry et al [1].

Fig. 2: Exam­ple scheme of a Knowl­edge-graph out­line queryable using SPARQL query lan­guage.

Sup­port from the 4TU.ResearchData FAIR DATA Fund

The FAIR DATA Fund enabled us to curate a knowl­edge-graph-based frame­work for inte­gra­tion of tran­scrip­tomics and metabolomics data gen­er­at­ed in this PhD project. Based on this frame­work, we are able to pro­vide inter­op­er­a­ble data for the plant nat­ur­al prod­ucts research com­mu­ni­ty suit­able for large-scale meta-analy­sis. We are able to com­ply with meta­da­ta stan­dards com­pat­i­ble with relat­ed omics data beyond the scope of this PhD project. With the sup­port pro­vid­ed by the FAIR DATA Fund, we are able to share the data gen­er­at­ed in this PhD project suit­able for re-use by a broad­er sci­en­tif­ic com­mu­ni­ty. In par­tic­u­lar, stan­dard­ized meta­da­ta curat­ed in this project aims to facil­i­tate biocu­ra­tion for AI devel­op­ment in the life sci­ences, as out­lined recent­ly [4].  

Ref­er­ences

[1] Gaudry, A., Pag­ni, M., Mehl, F., Moret­ti, S., Quiros-Guer­rero, L. M., Cap­pel­let­ti, L., … & Allard, P. M. (2024). A sam­ple-cen­tric and knowl­edge-dri­ven com­pu­ta­tion­al frame­work for nat­ur­al prod­ucts drug dis­cov­ery. DOI: 10.1021/acscentsci.3c00800.

[2] El Abiead, Y., Stro­bel, M., Payne, T., Fahy, E., O’Donovan, C., Sub­ra­mami­am, S., … & Wang, M. (2024). Enabling pan-repos­i­to­ry reanaly­sis for big data sci­ence of pub­lic metabolomics data. DOI: 10.26434/chemrxiv-2024-jt46s.

[3] Coop­er, L., Elser, J., Laporte, M. A., Arnaud, E., & Jaisw­al, P. (2024). Planteome 2024 Update: Ref­er­ence ontolo­gies and knowl­edge­base for plant biol­o­gy. Nucle­ic Acids Research, 52(D1), D1548-D1555. DOI: 10.1093/nar/gkad1028.

[4] Dessi­moz, C., & Thomas, P. D. (2024). AI and the democ­ra­ti­za­tion of knowl­edge. Sci­en­tif­ic Data, 11(1), 268. DOI: 10.1038/s41597-024–03099‑1

Feli­cia Wolters and the team at Wagenin­gen Uni­ver­si­ty & Research are among the win­ners of the FAIR Data Fund 2023 edi­tion.