Liang, G. | Search bioRxiv

bioRxiv2018

The rice G protein γ subunit qPE9-1 positively regulates grain-filling process by interacting with abscisic acid and auxin

The rice genome contains a single G (RGA1) and G{beta} (RGB1) and five G{gamma} subunits. Recent genetic studies have shown that DEP1/qPE9-1, an atypical putative G{gamma} protein, is responsible for dense and erect panicles, but the biochemical and molecular mechanisms underlying control of grain size are not well understood. Here, we report that plants carrying qPE9-1 have more endosperm cells per grain than plants contain the qpe9-1 allele. The qPE9-1 line has a higher rate and longer period of starch accumulation than the qpe9-1 line. Additionally, the expression of several key genes encoding enzymes catalyzing sucrose metabolism and starch biosynthesis is higher in the qPE9-1 line than in the qpe9-1 line, especially from the mid to late grain-filling stage. Grains of the qPE9-1 line also have higher contents of two phytohormones, ABA and IAA. Exogenous application of ABA or IAA enhanced starch accumulation and the expression of genes encoding grain-filling-related enzymes in the grains of qPE9-1, whereas only IAA produced these effects in qpe9-1. Based on these results, we conclude that qPE9-1 promotes endosperm cell proliferation and positively regulates starch accumulation largely through ABA and IAA, which enhance the expression of genes encoding starch biosynthesis during the late grain-filling stage.

plant biology↗

bioRxiv2018

GANai: Standardizing CT Images using Generative Adversarial Network with Alternative Improvement

Computed tomography (CT) is a widely-used diag-reproducibility regarding radiomic features, such as intensity, nostic image modality routinely used for assessing anatomical tissue characteristics. However, non-standardized imaging pro-tocols are commonplace, which poses a fundamental challenge in large-scale cross-center CT image analysis. One approach to address the problem is to standardize CT images using generative adversarial network models (GAN). GAN learns the data distribution of training images and generate synthesized images under the same distribution. However, existing GAN models are not directly applicable to this task mainly due to the lack of constraints on the mode of data to generate. Furthermore, they treat every image equally, but in real applications, some images are more difficult to standardize than the others. All these may lead to the lack-of-detail problem in CT image synthesis. We present a new GAN model called GANai to mitigate the differences in radiomic features across CT images captured using non-standard imaging protocols. Given source images, GANai composes new images by specifying a high-level goal that the image features of the synthesized images should be similar to those of the standard images. GANai introduces an alternative improvement training strategy to alternatively and steadily improve model performance. The new training strategy enables a series of technical improvements, including phase-specific loss functions, phase-specific training data, and the adoption of ensemble learning, leading to better model performance. The experimental results show that GANai is significantly better than the existing state-of-the-art image synthesis algorithms on CT image standardization. Also, it significantly improves the efficiency and stability of GAN model training.

bioinformatics↗