RiceMind Dataset

Abstract This repository contains the core datasets supporting the RiceMind platform, a comprehensive, structurally standardized, and fine-grained knowledge dataset for rice (Oryza sativa). To address the explosive growth of rice-related scientific literature, we employed automated, large-scale Natural Language Processing (NLP) strategies to mine experimentally validated associations from full-text articles (PubMed/PMC). This dataset not only provides extensive text-mined evidence for Gene-Trait Associations (GTAs), Gene-Variant Associations (GVAs), and Trait-Variant Associations (TVAs), but also seamlessly integrates these findings with multi-dimensional omics data and standardized ontologies. It serves as a robust resource for researchers in plant genomics, molecular breeding, and bioinformatics. Data Standardization & Identifier MappingTo ensure the highest level of structural consistency and cross-database interoperability, rigorous standardization pipelines were applied: Phenotypic Trait Standardization: All extracted trait descriptions have been systematically mapped and unified to recognized semantic ontologies, including the Gene Ontology (GO), Plant Trait Ontology (TO), Plant Ontology (PO), and the Rice Trait Ontology (RTO). The dataset centers around Oryza sativa. Gene Nomenclature Standardization: Gene entities sourced from external databases including Oryzabase, RAP-DB, Ensembl Plants, and Planteome were standardized to the unified RAP ID system. For instance, genes from Oryzabase were anchored via their annotated RAP IDs, and data from Planteome were mapped using a "Protein-Gene-RAP ID" trajectory to ensure absolute nomenclature consistency across the repository. Data Records Overview The repository includes the following files: 1. Text Corpus & Evidence Data keyword_filtered_rice_sentences.jsonl: Raw, keyword-filtered sentence segments extracted from full-text scientific literature. This serves as the foundational text corpus for downstream NLP processing. rice_context_sentences_compressed.tsv: Compressed contextual sentences providing the exact narrative evidence and semantic context from which the associations were mined. 2. NLP-Extracted Association Databases NLP_Rice_GTA_Database.tsv: The core text-mined Gene-Trait Association (GTA) dataset. It contains fine-grained, explicitly described relationships between rice genes and phenotypic traits extracted via automated NLP pipelines. NLP_Rice_GVA_TVA_Database.tsv: A supplementary association database capturing Gene-Variant Associations (GVAs) and Trait-Variant Associations (TVAs), mapping the genetic variations to specific traits and gene loci. NLP_Rice_GTA_Trend.tsv: Statistical and temporal trend data of the mined GTAs, reflecting the historical accumulation and rapid growth of specific rice breeding knowledge over time. 3. Integration & Standardization Masters Unified_Gene_Master.json: A comprehensive and standardized gene dictionary/ontology file. It resolves synonym conflicts and provides unified nomenclature for rice genes used across the entire dataset. rice_multi_omics.json: High-dimensional multi-omics data structures integrated to support and cross-validate the literature-mined narrative knowledge. Unified_Rice_GTA_Database.tsv: The definitive, ready-to-use master database. This file merges the NLP-extracted GTA dataset with the standardized gene master and multi-omics data layers, providing a fully integrated knowledge graph for downstream analysis. Usage Notes Files ending in .tsv are tab-separated values and can be opened with standard data science libraries (e.g., pandas in Python or readr in R). Files ending in .json and .jsonl contain structured hierarchical data suitable for programmatic parsing.