Long noncoding RNAs and mRNAs profiling in ovary during laying and broodiness in Taihe Black-Bone Silky Fowls (Gallus gallus Domesticus Brisson).

BACKGROUND: Broodiness significantly impacts poultry egg production, particularly notable in specific breeds such as the black-boned Silky, characterized by pronounced broodiness. An understanding of the alterations in ovarian signaling is essential for elucidating the mechanisms that influence broodiness. However, comparative research on the characteristics of long non-coding RNAs (lncRNAs) in the ovaries of broody chickens (BC) and high egg-laying chickens (GC) remains scant. In this investigation, we employed RNA sequencing to assess the ovarian transcriptomes, which include both lncRNAs and mRNAs, in eight Taihe Black-Bone Silky Fowls (TBsf), categorized into broody and high egg-laying groups. This study aims to provide a clearer understanding of the genetic underpinnings associated with broodiness and egg production. RESULTS: We have identified a total of 16,444 mRNAs and 18,756 lncRNAs, of which 349 mRNAs and 651 lncRNAs exhibited significantly different expression (DE) between the BC and GC groups. Furthermore, we have identified the cis-regulated and trans-regulated target genes of differentially abundant lncRNA transcripts and have constructed an lncRNA-mRNA trans-regulated interaction network linked to ovarian follicle development. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation analyses have revealed that DE mRNAs and the target genes of DE lncRNAs are associated with pathways including neuroactive ligand-receptor interaction, CCR6 chemokine receptor binding, G-protein coupled receptor binding, cytokine-cytokine receptor interaction, and ECM-receptor interaction. CONCLUSION: Our research presents a comprehensive compilation of lncRNAs and mRNAs linked to ovarian development. Additionally, it establishes a predictive interaction network involving differentially abundant lncRNAs and differentially expressed genes (DEGs) within TBsf. This significantly contributes to our understanding of the intricate interactions between lncRNAs and genes governing brooding behavior.

Identification and functional analysis of ovarian lncRNAs during different egg laying periods in Taihe Black-Bone Chickens.

Introduction: Long non-coding RNA (lncRNA) refers to a category of non-coding RNA molecules exceeding 200 nucleotides in length, which exerts a regulatory role in the context of ovarian development. There is a paucity of research examining the involvement of lncRNA in the regulation of ovary development in Taihe Black-Bone Chickens. In order to further investigate the egg laying regulation mechanisms of Taihe Black-Bone Chickens at different periods, transcriptome analysis was conducted on the ovarian tissues at different laying periods. Methods: This study randomly selected ovarian tissues from 12 chickens for RNA-seq. Four chickens were selected for each period, including the early laying period (102 days, Pre), the peak laying period (203 days, Peak), and the late laying period (394 days, Late). Based on our previous study of mRNA expression profiles in the same ovarian tissue, we identified three differentially expressed lncRNAs (DE lncRNAs) at different periods and searched for their cis- and trans-target genes to draw an lncRNA-mRNA network. Results and discussion: In three groups of ovarian tissues, we identified 136 DE lncRNAs, with 8 showing specific expression during the early laying period, 10 showing specific expression during the peak laying period, and 4 showing specific expression during the late laying period. The lncRNA-mRNA network revealed 16 pairs of lncRNA-target genes associated with 7 DE lncRNAs, and these 14 target genes were involved in the regulation of reproductive traits. Furthermore, these reproductive-related target genes were primarily associated with signaling pathways related to follicle and ovary development in Taihe Black-Bone Chickens, including cytokine-cytokine receptor interaction, TGF-beta signaling pathway, tyrosine metabolism, ECM-receptor interaction, focal adhesion, neuroactive ligand-receptor interaction, and cell adhesion molecules (CAMs). This study offers valuable insights for a comprehensive understanding of the influence of lncRNAs on poultry reproductive traits.

Transcriptomic and metabolomic analyses of the ovaries of Taihe black-bone silky fowls at the peak egg-laying and nesting period.

The poor reproductive performance of most local Chinese chickens limits the economic benefits and output of related enterprises. As an excellent local breed in China, Taihe black-bone silky fowl is in urgent need of our development and utilization. In this study, we performed transcriptomic and metabolomic analyses of the ovaries of Taihe black-bone silky fowls at the peak egg-laying period (PP) and nesting period (NP) to reveal the molecular mechanisms affecting reproductive performance. In the transcriptome, we identified five key differentially expressed genes (DEGs) that may affect the reproductive performance of Taihe black-bone silky fowl: BCHE, CCL5, SMOC1, CYTL1, and SCIN, as well as three important pathways: the extracellular region, Neuroactive ligand-receptor interaction and Cytokine-cytokine receptor interaction. In the metabolome, we predicted three important ovarian significantly differential metabolites (SDMs): LPC 20:4, Bisphenol A, and Cortisol. By integration analysis of transcriptome and metabolome, we identified three important metabolite-gene pairs: "LPC 20:4-BCHE", "Bisphenol A-SMOC1", and "Cortisol- SCIN". In summary, this study contributes to a deeper understanding of the regulatory mechanism of egg production in Taihe black-bone silky fowl and provides a scientific basis for improving the reproductive performance of Chinese local chickens.

Wearable Motion Capture: Reconstructing and Predicting 3D Human Poses From Wearable Sensors.

Reconstructing and predicting 3D human walking poses in unconstrained measurement environments have the potential to use for health monitoring systems for people with movement disabilities by assessing progression after treatments and providing information for assistive device controls. The latest pose estimation algorithms utilize motion capture systems, which capture data from IMU sensors and third-person view cameras. However, third-person views are not always possible for outpatients alone. Thus, we propose the wearable motion capture problem of reconstructing and predicting 3D human poses from the wearable IMU sensors and wearable cameras, which aids clinicians' diagnoses on patients out of clinics. To solve this problem, we introduce a novel Attention-Oriented Recurrent Neural Network (AttRNet) that contains a sensor-wise attention-oriented recurrent encoder, a reconstruction module, and a dynamic temporal attention-oriented recurrent decoder, to reconstruct the 3D human pose over time and predict the 3D human poses at the following time steps. To evaluate our approach, we collected a new WearableMotionCapture dataset using wearable IMUs and wearable video cameras, along with the musculoskeletal joint angle ground truth. The proposed AttRNet shows high accuracy on the new lower-limb WearableMotionCapture dataset, and it also outperforms the state-of-the-art methods on two public full-body pose datasets: DIP-IMU and TotalCaputre.

Association of variants and expression levels of MYOD1 gene with carcass and muscle characteristic traits in domestic pigeons.

This study was to investigate the correlations of myogenic differentiation 1 (MYOD1) gene polymorphisms with carcass traits and its expression with breast muscle development in pigeons. Four SNPs were found in the pigeon MYOD1 gene. Correlation analysis showed that individuals with AA genotype at both SNPs g.2967A > G (p < .01) and g.3044G > A (p < .05) have significantly higher live weight (LW), carcass weight (CW), semi-eviscerated weight (SEW), eviscerated weight (EW) and breast muscle weight (BMW). Moreover, the two SNPs also had the same significant effects on MYOD1 mRNA expression levels in breast muscle of pigeons, ie, the AA genotype showed higher MYOD1 mRNA expression levels. The diameter and cross-section area of muscle fibers continuously increased from 0w to 4w (p < .05), accompanied with the increasing expression of MYOD1 gene, while the density decreased (p < .05) dramatically from 0w to 1w and continuously fell over in the next few weeks (p > .05). What's more, the expression level of MYOD1 gene was positively correlated with a diameter (r = 0.937, p < .05) and cross-sectional area (r = 0.956, p < .01) of myofiber, and negatively correlated with density (r = -0.769, p < .01). The results showed that individuals with AA genotype at both SNPs g.2967A > G and g.3044G > A have showed higher carcass traits (LW, CW, SEW, EW, and BMW) and higher MYOD1 mRNA expression level in breast muscle than AB and BB genotypes. Moreover, the expression level of MYOD1 gene was closely correlated with muscle characteristic traits, indicating variants of MYOD1 gene was closely related to muscle development and could be a potential candidate gene in marker-assisted selection of pigeons.

Phase Contrast Image Restoration by Formulating Its Imaging Principle and Reversing the Formulation With Deep Neural Networks.

Phase contrast microscopy, as a noninvasive imaging technique, has been widely used to monitor the behavior of transparent cells without staining or altering them. Due to the optical principle of the specifically-designed microscope, phase contrast microscopy images contain artifacts such as halo and shade-off which hinder the cell segmentation and detection tasks. Some previous works developed simplified computational imaging models for phase contrast microscopes by linear approximations and convolutions. The approximated models do not exactly reflect the imaging principle of the phase contrast microscope and accordingly the image restoration by solving the corresponding deconvolution process is not perfect. In this paper, we revisit the optical principle of the phase contrast microscope to precisely formulate its imaging model without any approximation. Based on this model, we propose an image restoration procedure by reversing this imaging model with a deep neural network, instead of mathematically deriving the inverse operator of the model which is technically impossible. Extensive experiments are conducted to demonstrate the superiority of the newly derived phase contrast microscopy imaging model and the power of the deep neural network on modeling the inverse imaging procedure. Moreover, the restored images enable that high quality cell segmentation task can be easily achieved by simply thresholding methods. Implementations of this work are publicly available at https://github.com/LiangHann/Phase-Contrast-Microscopy-Image-Restoration.

Transcriptome Analysis of the Ovaries of Taihe Black-Bone Silky Fowls at Different Egg-Laying Stages.

The poor egg-laying performance and short peak egg-laying period restrict the economic benefits of enterprises relating to the Taihe black-bone silky fowl. Ovaries are the main organ for egg production in poultry. Unlike that of mammals, the spawning mechanism of poultry has rarely been reported. As a prominent local breed in China, the reproductive performance of Taihe black-bone silky fowls is in urgent need of development and exploitation. To further explore the egg-laying regulation mechanism in the different periods of Taihe black-bone silky fowls, the ovarian tissues from 12 chickens were randomly selected for transcriptome analysis, and 4 chickens in each of the three periods (i.e., the pre-laying period (102 days old, Pre), peak laying period (203 days old, Peak), and late laying period (394 days old, Late)). A total of 12 gene libraries were constructed, and a total of 9897 differential expression genes (DEGs) were identified from three comparisons; the late vs. peak stage had 509 DEGs, the pre vs. late stage had 5467 DEGs, and the pre vs. peak stage had 3921 DEGs (pre-stage: pre-egg-laying period (102 days old), peak-stage: peak egg-laying period (203 days old), and late-stage: late egg-laying period (394 days old)). In each of the two comparisons, 174, 84, and 2752 differentially co-expressed genes were obtained, respectively, and 43 differentially co-expressed genes were obtained in the three comparisons. Through the analysis of the differential genes, we identified some important genes and pathways that would affect reproductive performance and ovarian development. The differential genes were LPAR3, AvBD1, SMOC1, IGFBP1, ADCY8, GDF9, PTK2B, PGR, and CD44, and the important signaling pathways included proteolysis, extracellular matrices, vascular smooth muscle contraction, the NOD-like receptor signaling pathway and the phagosome. Through the analysis of the FPKM (Fragments Per Kilobase of exon model per Million mapped fragments) values of the genes, we screened three peak egg-laying period-specific expressed genes: IHH, INHA, and CYP19A1. The twelve genes and five signaling pathways mentioned above have rarely been reported in poultry ovary studies, and our study provides a scientific basis for the improvement of the reproductive performance in Taihe black-bone silky fowls.

Corrigendum: Expression of DGAT2 gene and its associations with intramuscular fat content and breast muscle fiber characteristics in domestic pigeons (Columba livia).

[This corrects the article DOI: 10.3389/fvets.2022.847363.].

Transcriptomics and Metabolomics Analysis of the Ovaries of High and Low Egg Production Chickens.

Egg production is a pivotal indicator for evaluating the fertility of poultry, and the ovary is an essential organ for egg production and plays an indispensable role in poultry production and reproduction. In order to investigate different aspects of egg production mechanisms in different poultry, in this study we performed a metabolomic analysis of the transcriptomic combination of the ovaries of two chicken breeds, the high-production Ninghai indigenous chickens and the low-production Wuliangshan black-boned chickens, to analyze the biosynthesis and potential key genes and metabolic pathways in the ovaries during egg production. We predicted four genes in the transcriptomic that are associated with egg production, namely P2RX1, INHBB, VIPR2, and FABP3, and identified three important pathways during egg production, "Calcium signaling pathway", "Neuroactive ligand-receptor interaction" and "Cytokine-cytokine receptor interaction", respectively. In the metabolomic 149 significantly differential metabolites were identified, 99 in the negative model and 50 in the positive model, of which 17α-hydroxyprogesterone, iloprost, spermidine, and adenosine are important metabolites involved in reproduction. By integrating transcriptomics and metabolomics, the correlation between specific differential genes and differential metabolites identified important gene-metabolite pairs "VIPR2-Spermidine" and "P2RX1-Spermidine" in egg production. In conclusion, these data provide a better understanding of the molecular differences between the ovaries of low- and high-production hens and provide a theoretical basis for further studies on the mechanics of poultry egg production.

Expression of DGAT2 Gene and Its Associations With Intramuscular Fat Content and Breast Muscle Fiber Characteristics in Domestic Pigeons (Columba livia).

Diacylglycerol acyltransferase 2 (DGAT2) catalyzes the final step in triglyceride synthesis and plays an important role in the synthesis of fat, but the effects of its expression on intramuscular fat (IMF) content and muscle development are still unknown. In this study, we investigated the expression of the DGAT2 gene and its associations with IMF content and breast muscle fiber characteristics in pigeons. The spatiotemporal expression profile of the pigeon DGAT2 gene in breast muscle showed that the mRNA expression level of DGAT2 gene in subcutaneous fat was the highest (p < 0.01) among eight tissues from 0 to 4 weeks of age, and showed an upward trend week by week, followed by liver (p < 0.05). Moreover, both mRNA and protein levels of the DGAT2 gene in breast muscle showed an upward trend from 0 to 4 weeks (p < 0.05), accompanied by the upregulation of MYOD1 and MSTN. In addition, the paraffin section analysis results revealed that the diameter and cross-sectional area of pectoralis muscle fiber significantly increased with age (p < 0.05), and a significant positive correlation was shown between the DGAT2 gene expression level and muscle fiber diameter (p < 0.05). Furthermore, correlation analysis suggested that the mRNA expression level of the pigeon DGAT2 gene was significantly (p < 0.01) correlated with IMF content in breast muscle. These results imply that the DGAT2 gene has a close relationship with IMF content and breast muscle fiber characteristics in pigeons, indicating that the DGAT2 gene might be used as a candidate gene marker-assisted breeding in pigeons.

Identification of Genes Related to Squab Muscle Growth and Lipid Metabolism from Transcriptome Profiles of Breast Muscle and Liver in Domestic Pigeon (Columba livia).

The improvements in muscle growth rate and meat quality are the major breeding aims in pigeon industry. Liver and muscle are recognized as important sites for fatty acid metabolism; understanding the role of specific transcripts in the breast muscle and liver might lead to the elucidation of interrelated biological processes. In this study, RNA-Sequencing (RNA-Seq) was applied to compare the transcriptomes of breast muscle and liver tissues among pigeons at five developmental periods (0, 1, 2, 3, 4 weeks post-hatching) to identify candidate genes related to muscle growth and lipid metabolism. There were 3142 differentially expressed genes (DEGs) identified in the breast muscle libraries; 1794 genes were up-regulated while 1531 genes were down-regulated. A total of 1323 DEGs were acquired from the liver libraries, with 791 up-regulated genes and 591 down-regulated genes. By pathway enrichment analysis, a set of significantly enriched pathways were identified for the DEGs, which are potentially involved in cell proliferation and differentiation, lipid metabolism and energy metabolism in pigeon breast muscle and liver. Our results are consistent with previous partial reports from domestic animals and poultry and provide some unidentified genes involved in muscle growth and lipid metabolism. The reliability of the sequencing data was verified through qPCR analysis of 16 genes from eight comparison groups (two genes per group). The findings from this study could contribute to future investigations of muscle growth and lipid metabolism mechanisms and establish molecular approaches to improve muscle growth rate and meat quality in domestic pigeon breeding.

Ovarian Transcriptomic Analysis of Ninghai Indigenous Chickens at Different Egg-Laying Periods.

Egg production is an essential indicator of poultry fertility. The ovary is a crucial organ involved in egg production; however, little is known about the key genes and signaling pathways involved in the whole egg-laying cycle of hens. In order to explore the mechanism of egg production at different stages of the egg-laying process, ovarian tissues from four chickens were randomly selected for transcriptome analysis at each of the three ages (145 d, 204 d, and 300 d in the early, peak, and late stages of egg laying). A total of 12 gene libraries were constructed, and a total of 8433 differential genes were identified from NH145d vs. NH204d, NH145d vs. NH300d and NH300d vs. NH204d (Ninghai 145-day-old, Ninghai 204-day-old, and Ninghai 300-day-old), with 1176, 1653 and 1868 up-regulated genes, and 621, 1955 and 1160 down-regulated genes, respectively. In each of the two comparison groups, 73, 1004, and 1030 differentially expressed genes were found to be co-expressed. We analyzed the differentially expressed genes and predicted nine genes involved in egg production regulation, including LRP8, BMP6, ZP4, COL4A1, VCAN, INHBA, LOX, PTX3, and IHH, as well as several essential egg production pathways, such as regulation adhesion molecules (CAMs), calcium signaling pathways, neuroactive ligand-receptor interaction, and cytokine-cytokine receptor interaction. Transcriptional analysis of the chicken ovary during different phases of egg-lay will provide a useful molecular basis for study of the development of the egg-laying ovary.

Investigation of the CEBPA gene expression pattern and association analysis of its polymorphisms with meat quality traits in chickens.

Meat quality is closely related to the fat deposition which is regulated by a cascade of transcription factors. As a transcription factor, the CCAAT/enhancer binding protein alpha (CEBPA) is considered as one of the key molecules regulating adipogenesis. Therefore, the objective of this study was to detect the expression pattern of the CEBPA gene and evaluate whether its single nucleotide polymorphisms (SNPs) were associated with the meat quality traits in Wuliang Mountain Black-bone (WLMB) chickens. The results showed that the chicken CEBPA mRNA was widely expressed in the 11 tissues, and the expression pattern of it might be tissue- and time-specific different. The locus of g.74C > G was not significantly associated with chicken meat quality. For the locus of g.552G > A, chickens with the GG genotype showed higher pH (p < 0.01), lower drip loss (p < 0.01) and higher intramuscular fat (p < 0.05) than those with other genotypes. It suggested that polymorphisms of the CEBPA gene were significantly associated with the meat quality traits of WLMB chickens. The results of this study contribute to the functional research of the CEBPA gene and lay the foundation for improving meat quality based on the marker-assisted selection in chickens.

A survey on applications of deep learning in microscopy image analysis.

Advanced microscopy enables us to acquire quantities of time-lapse images to visualize the dynamic characteristics of tissues, cells or molecules. Microscopy images typically vary in signal-to-noise ratios and include a wealth of information which require multiple parameters and time-consuming iterative algorithms for processing. Precise analysis and statistical quantification are often needed for the understanding of the biological mechanisms underlying these dynamic image sequences, which has become a big challenge in the field. As deep learning technologies develop quickly, they have been applied in bioimage processing more and more frequently. Novel deep learning models based on convolution neural networks have been developed and illustrated to achieve inspiring outcomes. This review article introduces the applications of deep learning algorithms in microscopy image analysis, which include image classification, region segmentation, object tracking and super-resolution reconstruction. We also discuss the drawbacks of existing deep learning-based methods, especially on the challenges of training datasets acquisition and evaluation, and propose the potential solutions. Furthermore, the latest development of augmented intelligent microscopy that based on deep learning technology may lead to revolution in biomedical research.

Comparative Transcriptome Profiling of mRNA and lncRNA of Ovaries in High and Low Egg Production Performance in Domestic Pigeons (Columba livia).

Egg production performance is one of the most important economic traits in pigeon industry. However, little is known regarding how egg production performance is regulated by long non-coding RNAs (lncRNAs) in pigeons. To evaluate the lncRNAs and mRNAs in ovaries associated with egg production performance in domestic pigeons, high-throughput RNA sequencing of ovaries between high and low egg production performance groups were performed and analyzed in this study. A total of 34,346 mRNAs and 24,601 lncRNAs were identified, including 14,525 known lncRNAs and 10,076 novel lncRNAs, of which 811 mRNAs and 148 lncRNAs (P < 0.05) were significantly differentially expressed (DE) between the groups of high and low egg production performance. GO and KEGG annotation analysis indicated that the target genes of DE lncRNAs and DE mRNAs were related to cell differentiation, ATP binding and methylation. Moreover, we found that FOXK2, a target gene of lncRNA MSTRG.7894.4, was involved in regulating estrogen receptors. Our study provided a catalog of lncRNAs and mRNAs associated with egg production performance, and they deserve further study to deepen the understanding of biological processes in the ovaries of pigeons.

Chronological Expression of PITX2 and SIX1 Genes and the Association between Their Polymorphisms and Chicken Meat Quality Traits.

Meat quality is closely related to the development of skeletal muscle, in which PITX2 and SIX1 genes play important regulatory roles. The present study firstly provided the data of chronological expression files of PITX2 and SIX1 genes in the post-hatching pectoral muscle and analyzed the association of their polymorphisms with the meat quality traits of Wuliang Mountain Black-bone (WLMB) chickens. The results showed that both PITX2 and SIX1 genes were weakly expressed in the second and third weeks, and then increased significantly from the third week to the fourth week. Furthermore, there was a significant positive correlation between the expression levels of the two genes. Twelve and one SNPs were detected in the chicken PITX2 and SIX1 genes, respectively, of which four SNPs (g.9830C > T, g.10073C > T, g.13335G > A, g.13726A > G) of the PITX2 gene and one SNP (g.564G > A) of the SIX1 gene were significantly associated with chicken meat quality traits. For the PITX2 gene, chickens with the CT genotype of g.9830C > T showed the highest meat color L*, shear force (SF), pH, and the lowest electrical conductivity (EC), and drip loss (DL) (p < 0.05 or p < 0.01); chickens with the CC genotype of g.10073C > T had the lowest L*, pH, and the highest DL (p < 0.01). For the SIX1 gene, chickens with the GG genotype of g.564G > A had the highest (p < 0.05) SF and pH. Furthermore, pH had a significant correlation with all the other meat quality traits. The current study could contribute to the research of regulatory mechanisms of meat quality and lay the foundation for improving meat quality based on marker-assisted selection in chickens.

Crash report data analysis for creating scenario-wise, spatio-temporal attention guidance to support computer vision-based perception of fatal crash risks.

Reducing traffic fatal crashes has been an important mission of transportation. With the rapid development of sensor and Artificial Intelligence (AI) technologies, the computer vision (CV)-based crash anticipation in the near-crash phase is receiving growing attention. The ability to perceive fatal crash risks in an early stage is of paramount importance as well because it can improve the reliability of crash anticipation. Yet this task is challenging because it requires establishing a relationship between the driving scene information that CV can recognize and the fatal crash features that CV will not get until the crash occurrence. Image data with the annotation for directly training a reliable AI model for the early visual perception of fatal crash risks are not abundant. The Fatality Analysis Reporting System (FARS) contains big data on fatal crashes, which is a reliable data source for finding fatal crash clusters and discovering their distribution patterns to tell the association between driving scene characteristics and fatal crash features. To enhance CV's ability to perceive fatal crash risks earlier, this paper develops a data analytics model from fatal crash report data, which is named scenario-wise, spatio-temporal attention guidance. First, the paper identifies five descriptive variables that are sparse and thus allow for decomposing the 5-year (2013-2017) fatal crash dataset to develop scenario-wise attention guidance. Then, an exploratory analysis of location- and time-related descriptive variables suggests dividing fatal crashes into spatially defined groups. A group's temporal distribution pattern is an indicator of the similarity of fatal crashes in the group. Hierarchical clustering and K-means clustering further merge the spatially defined groups into six clusters according to the similarity of their temporal patterns. After that, association rule mining discovers the statistical relationship between the temporal information of driving scenes with fatal crash features, such as the first harmful event and the manner of collisions, for each cluster. The paper illustrates how the developed attention guidance supports the design and implementation of a preliminary CV model that can identify agents of a possibility to involve in fatal crashes from their environmental and context information.

Expression analysis of microRNAs and their target mRNAs of testes with high and low sperm motility in domestic pigeons (Columba livia).

Sperm motility is one of the most important indicators to evaluate poultry fertility. In order to explore key molecular regulation roles related to sperm motility, we employed testicular RNA sequencing of pigeon. A total of 705 known and 385 novel microRNAs were identified. Compared with the low sperm motility group, four upregulated and two downregulated miRNAs in the high sperm motility group were identified. A total of 3567 target mRNAs were predicted and four target mRNAs were selected to validate by qPCR. The miRNA-mRNA interaction network analysis, indicated that mmu-miR-183-5p /FOXO1 and PC-3p-244994_31/CHDH pairs might affect sperm motility. GO and KEGG annotation analysis showed that target genes of differentially expressed miRNAs were related to serine/threonine kinase activity, ATP binding, Wnt and MAPK signaling pathway. The study provided a global miRNAs transcriptome of pigeon and a novel insight into the expression of the miRNAs in testes that associated with sperm motility.

Weakly Supervised Cell Segmentation by Point Annotation.

We propose weakly supervised training schemes to train end-to-end cell segmentation networks that only require a single point annotation per cell as the training label and generate a high-quality segmentation mask close to those fully supervised methods using mask annotation on cells. Three training schemes are investigated to train cell segmentation networks, using the point annotation. First, self-training is performed to learn additional information near the annotated points. Next, co-training is applied to learn more cell regions using multiple networks that supervise each other. Finally, a hybrid-training scheme is proposed to leverage the advantages of both self-training and co-training. During the training process, we propose a divergence loss to avoid the overfitting and a consistency loss to enforce the consensus among multiple co-trained networks. Furthermore, we propose weakly supervised learning with human in the loop, aiming at achieving high segmentation accuracy and annotation efficiency simultaneously. Evaluated on two benchmark datasets, our proposal achieves high-quality cell segmentation results comparable to the fully supervised methods, but with much less amount of human annotation effort.

Detecting Medical Misinformation on Social Media Using Multimodal Deep Learning.

In 2019, outbreaks of vaccine-preventable diseases reached the highest number in the US since 1992. Medical misinformation, such as antivaccine content propagating through social media, is associated with increases in vaccine delay and refusal. Our overall goal is to develop an automatic detector for antivaccine messages to counteract the negative impact that antivaccine messages have on the public health. Very few extant detection systems have considered multimodality of social media posts (images, texts, and hashtags), and instead focus on textual components, despite the rapid growth of photo-sharing applications (e.g., Instagram). As a result, existing systems are not sufficient for detecting antivaccine messages with heavy visual components (e.g., images) posted on these newer platforms. To solve this problem, we propose a deep learning network that leverages both visual and textual information. A new semantic- and task-level attention mechanism was created to help our model to focus on the essential contents of a post that signal antivaccine messages. The proposed model, which consists of three branches, can generate comprehensive fused features for predictions. Moreover, an ensemble method is proposed to further improve the final prediction accuracy. To evaluate the proposed model's performance, a real-world social media dataset that consists of more than 30,000 samples was collected from Instagram between January 2016 and October 2019. Our 30 experiment results demonstrate that the final network achieves above 97% testing accuracy and outperforms other relevant models, demonstrating that it can detect a large amount of antivaccine messages posted daily. The implementation code is available at https://github.com/wzhings/antivaccine_detection.