Gene survey for sustainable rice production with low-fertilizer and establishment of a new phenotyping system applying deep learning.

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K05529
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 39010:Science in plant genetics and breeding-related
• Research Institution: The University of Shiga Prefecture
• Principal Investigator: Shimizu Akifumi 滋賀県立大学, 環境科学部, 准教授 (40409082)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 遺伝育種学 / 栄養ストレス耐性

Outline of Final Research Achievements

We produced 188 recombinant inbred lines of the standard rice cultivar Nipponbare and the nutrient stress-tolerant cultivar KHAO NOK, and we obtained high-density genotype information across the entire genome using the GRAS-Di method. These lines were grown hydroponically under low and no nutrition conditions, and QTL analysis was conducted on dry weight. It was found that the accumulation of multiple QTLs could explain the difference between the varieties. Currently, multi-element analysis using ICP-MS is being conducted to isolate the detected QTLs.
In preparation for trait classification using deep learning, we used a population of hybrid offspring with separated ear traits and we advanced a convolutional neural network with the separated marker genotype as the teacher label.

Free Research Field

遺伝育種科学

Academic Significance and Societal Importance of the Research Achievements

低栄養処理区特異的に検出されたQTLsは、40倍希釈した水耕液という低濃度の養分を効率的に利用してバイオマスを増加する機能を持ちうる。これらの利用は、投入肥料を削減することで持続的な農業の実現に役立つといえる。GRAS-Di法で得た日本晴×KHAO NOKの組換え自殖後代のゲノム全域に渡る高密度遺伝子型情報は、上記の低栄養特異的QTLsの検出に有効なだけでなく、親品種間に観られている様々な栄養ストレス耐性遺伝子座の検出にも利用できる。また、得られた高密度遺伝子型情報を深層学習時の教師ラベルとして活用することで、単純な分類が不可能だったメンデル遺伝する新しい表現型値の探索法の確立に有効である。