关键词: 自动化平台, 深度学习, 克隆挑选, 机器视觉, 合成生物学

Abstract:

Microbial clone picking, a crucial step in genetic engineering and biological experiments, involves the accurate and rapid isolation of single colonies with desired characteristics from petri dishes teeming with numerous clones, followed by their inoculation into culture media for further propagation or analysis. In high-throughput settings, this task becomes burdensome due to its vast volume, complex record-keeping requirements, and the risk of cross-contamination, rendering manual operations impractical for achieving timely and precise results. To address this challenge, we present the design and manufacture of an automated clone picking workstation that performs efficient clone picking using 96-channel pneumatic pick-up pins, eliminating the need for consumables. The pins can be reused after ultrasonic cleaning and sterilization at high temperature following the previous picking cycle, making it more economical and environmentally friendly, compared with other methods that use disposable pipettes or picking needles. The pins can be replaced to adapt to different types of bacterial strains to meet various experimental requirements. The grab integrated on the picking head can rotate 360° and transfer the plates to different work positions.In the aspect of colony detection, the photos are automatically taken by an optical system, and the positioning and screening of colonies are achieved through the deep learning of numerous colony images by the software, which was designed independently. The precision image recognition technology is coupled with robotic and automated control technologies to enable seamless processes for picking, inoculating, cleaning, and drying. The High-Efficiency Particulate Air Filter and ultraviolet sterilization prevent cross-contamination, ensuring the experimental environment meets the required standards. This workstation is equipped with an independent operation computer and has developed a set of user-friendly software that enables personalized editing of multiple experimental protocols tailored to diverse microbial clone types. It can also communicate with external devices via TCP/IP protocols, facilitating the integration for conducting experiments such as fully automated synthetic biology. The validation experiment of bacterial colony picking was conducted by a prototype machine to test the selection efficiency. The success of the experiment suggests that the proposed system and method are feasible and effective, offering a valuable tool and a practical approach for the automation development of high-throughput laboratories. {L-End}

Key words: automation platform, deep learning, clone picking, machine vision, synthetic biology