What is GREGOR?
GREGOR is an AI-assisted platform for plant trait genomics. It provides an integrated pipeline that takes researchers from scientific literature all the way to a ranked list of candidate genes in their organism of interest — supporting both gene editing target selection and broader biological process exploration.
GREGOR is designed for plant biologists, genomic selection researchers, and crop developers working with model and non-model plant species. It integrates text mining, orthology mapping, gene regulatory network analysis, and genomic annotation into a single collaborative environment.
Open the User ManualTrait Design & Gene Discovery
GREGOR's main module. It turns a trait term as it appears in the literature into a prioritized list of candidate genes, backed by the evidence behind each one. You name the trait, pick the organism and tissues, add your own expression data, and GREGOR walks the pipeline: text-mining of millions of PubMed abstracts, MENDEL reading every abstract to keep only real gene-trait evidence, orthology mapping onto your genome, and the regulatory network that adds the genes no paper has published yet.
Any term used in the scientific literature works as a starting point: a trait, a phenotype, a biological process, a stress condition. You can combine several, and MENDEL tells you which associations it kept and which it discarded, and why.
MENDEL
The AI assistant inside GREGOR
Text-mining returns every gene that appears near your trait in the literature, and most of those are coincidences: two names in the same abstract for unrelated reasons. MENDEL reads each abstract and keeps only the ones that state a real gene-trait relationship.
It doesn't just filter, it explains: for every gene it kept or discarded you get the reason and the sentence it came from, so the decision is yours to review, not a black box to trust.
The analysis runs in one of three modes, which changes how candidates are ranked
Gene Discovery
Find genes associated with the trait without a predefined editing goal. Useful to understand trait architecture, spot key regulators and generate hypotheses.
Repression
Rank the analysis toward genes best suited to silencing (knockout / knockdown), based on their position in the regulatory network.
Overexpression
Rank the analysis toward genes best suited to overexpression, for traits you want to amplify rather than remove.
Text Mining — Mining scientific literature
Automatically mines thousands of PubMed articles to identify genes associated with your trait or biological process of interest. GREGOR surfaces the evidence from published research.
Orthology — From literature to your organism
Genes found in literature often come from model species like Arabidopsis. GREGOR maps them directly to your target organism's genome via NCBI RefSeq orthology — so you work with the exact genes in your species of interest, not just surrogates.
GRN — The regulatory network of your trait
GREGOR constructs the Gene Regulatory Network (GRN) of your trait, revealing functional interactions and regulatory relationships between genes. Genes with lower network centrality often represent the most promising intervention targets, with more specific effects and lower pleiotropy risk.
Candidate Genes — Ranked and ready to act
A prioritized list of candidate genes ranked by integrated evidence from literature mining, orthology mapping, network centrality, and tissue-specific expression patterns. Results are designed for collaborative review, enabling multi-user annotation, validation, and selection.
Gene Editing
A complementary module: the pipeline returns a ranked list of candidate genes, and the CRISPR Workspace is what turns any of them into editing constructs, when that's where your work is going. The genes you select in the candidate list arrive already grouped, and the workspace is embedded in the pipeline itself, showing only the current candidates and which of them already have a design.
It also works on its own. Search & Edit reaches every gene of every organism in the database, designed or not, so a construct doesn't have to start from a trait design. Reached from Resources → Gene Editing, or from the gene information panel of any gene.
From a set of candidates to a set of guides
Batch design
Pick as many genes as you want and design them in one batch, choosing the mutation type, the nuclease and the deaminase: seven nucleases (Cas9, Cas12, CasPhi, SauriCas9, CjCas9, Nme2Cas9, dTnpB) and fifteen deaminases. The work runs in the background and reports as each design lands, so the tab can be closed without losing anything.
Guides and off-targets
Every design opens guide by guide, with its GC content and its off-targets scored across the genome. For base editors you also get the window it edits, the sequence it produces and the amino acid it changes. Each guide ID is searchable in the genome browser, so you can see exactly where it lands, down to the base.
Traceability
The same gene takes as many designs as you want, side by side: another nuclease, another set of criteria, and the guides change with them. Designs group by organism, by trait, by gene set, or by the trait design they came from, which links back to it.
Explore Genes
Don't have a trait design in mind? Explore the database directly. Six search modes reach every annotated gene across the supported organisms, and every result opens the same evidence panels: orthologs, regulatory network, annotations and expression.
Search by Name / ID
Find genes by their identifier or name across all supported organisms and assemblies. Optionally extend the search to gene descriptions for broader results.
Search by Annotation
The most powerful search mode in GREGOR. Annotations are generated by a proprietary deep learning model, enabling accurate functional annotation across all supported organisms — including non-model species that lack well-curated reference annotations.
Unlike databases that rely on annotation transfer from model organisms like Arabidopsis thaliana, GREGOR's annotations are computed directly for each genome, ensuring reliability even for newly assembled or understudied species.
Supported annotation vocabularies
Search by Organism
Browse all genes within a specific plant species or assembly. Explore genomic diversity and access the complete gene catalog for any supported organism.
Search by Sequence (BLAST)
Submit an amino acid or nucleotide sequence and identify matching genes across the full GREGOR database or within a specific organism using BLAST-based alignment. Ideal for identifying genes from experimental data or cross-species comparisons.
Search by Networks
Give GREGOR several genes at once and it draws the regulatory network they share. Each edge is a source and a target, scored by three independent inference methods, and the network can be widened to the second degree or filtered by annotation and expression.
Search by Expression
List every gene expressed in a tissue above a TPM threshold you set, narrowed by developmental stage where the organism has it. Built on the same tissue expression data the pipeline uses to rank candidates.
Supported Organisms
These species are enabled for analysis today:
Ten more plant genomes are already annotated in the database and can be enabled on request, including peach, sweet cherry, rice, grapevine, watermelon and several citrus species. Which organisms your account can analyse depends on your plan.
Need a species that isn't here? Contact us at gregor@meristem.bio to discuss adding your genome of interest.
The Team
How to Cite
If you use GREGOR in your research, please cite:
GREGOR v3. Meristem.bio. https://gregor.meristem.bio
Contact
For questions, access requests, or collaboration inquiries: