Creator 1.0: Phylogeny-Guided Stagewise Pretraining for Plant Genome Foundation Models across 1,000 Species
- 1Hainan University
Abstract
Genomic language models have advanced DNA sequence analysis, but their applicability to plant-specific tasks and representation changes during pretraining remain poorly characterized. Here, we present Creator 1.0, a 2-billion-parameter parameter plant genome foundation model pretrained on 1,000 plant genomes using phylogeny-guided stagewise pretraining (PGSP). Training comprised 83 consecutive stages and produced 52 cumulative checkpoints associated with distinct plant orders. On GENEB, the stage 83 checkpoint ranked first among 46 models across 10 plant-related tasks, with a mean Matthews correlation coefficient (MCC) of 0.565, and fourth across all 100 tasks. Evaluation of 52 checkpoints on six plant lncRNA tasks identified different optimal stages under full-data, 10-shot and 1-shot settings, corresponding to stages 71, 72 and 29, respectively. Checkpoint performance varied across tasks, without consistent advantages for corresponding plant orders. These findings demonstrate the competitive performance of Creator 1.0 and highlight the potential of PGSP to provide task-adaptable checkpoints for plant genomic applications.
Related
- Re-evaluating the benefit of pretraining scale in cell foundation models
- Phylogeny and Species Delimitation in Isoxylosteum, a Lonicera Clade Endemic to the Himalayan-Tibetan-Hengduan Region
- Phylogeny and comparative chloroplast genomics reveal a rare new short-styled homostylous species of Primula (Primulaceae) from southwestern China
- BarcodeMAE+: Rethinking Masked Pretraining and Global Representations for DNA Barcode Foundation Models
- Phylogeny and reassessment of the genus Tolypocladium s. lato.: Proposing a new genus, three new species and two combinations.
BibTeX
@misc{yang2026creator,
title = {{Creator 1.0: Phylogeny-Guided Stagewise Pretraining for Plant Genome Foundation Models across 1,000 Species}},
author = {L. Yang and S. Jiang and Y. Li and M. Zou and Y. Xia and L. Wang and Y. Wu and H. Wang and W. Xie and Z. Xia},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.22.761197},
url = {https://www.biorxiv.org/content/10.64898/2026.10.22.761197v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Creator 1.0: Phylogeny-Guided Stagewise Pretraining for Plant Genome Foundation Models across 1,000 Species
AU - L. Yang
AU - S. Jiang
AU - Y. Li
AU - M. Zou
AU - Y. Xia
AU - L. Wang
AU - Y. Wu
AU - H. Wang
AU - W. Xie
AU - Z. Xia
PY - 2026
DA - 2026/10/22
PB - bioRxiv
DO - 10.64898/2026.10.22.761197
UR - https://www.biorxiv.org/content/10.64898/2026.10.22.761197v1
AB - Genomic language models have advanced DNA sequence analysis, but their applicability to plant-specific tasks and representation changes during pretraining remain poorly characterized. Here, we present Creator 1.0, a 2-billion-parameter parameter plant genome foundation model pretrained on 1,000 plant genomes using phylogeny-guided stagewise pretraining (PGSP). Training comprised 83 consecutive stages and produced 52 cumulative checkpoints associated with distinct plant orders. On GENEB, the stage 83 checkpoint ranked first among 46 models across 10 plant-related tasks, with a mean Matthews correlation coefficient (MCC) of 0.565, and fourth across all 100 tasks. Evaluation of 52 checkpoints on six plant lncRNA tasks identified different optimal stages under full-data, 10-shot and 1-shot settings, corresponding to stages 71, 72 and 29, respectively. Checkpoint performance varied across tasks, without consistent advantages for corresponding plant orders. These findings demonstrate the competitive performance of Creator 1.0 and highlight the potential of PGSP to provide task-adaptable checkpoints for plant genomic applications.
ER -