Positive-unlabeled learning in bioinformatics and computational biology: a brief review.

Conventional supervised binary classification algorithms have been widely applied to address significant research questions using biological and biomedical data. This classification scheme requires two fully labeled classes of data (e.g. positive and negative samples) to train a classification model. However, in many bioinformatics applications, labeling data is laborious, and the negative samples might be potentially mislabeled due to the limited sensitivity of the experimental equipment. The positive unlabeled (PU) learning scheme was therefore proposed to enable the classifier to learn directly from limited positive samples and a large number of unlabeled samples (i.e. a mixture of positive or negative samples). To date, several PU learning algorithms have been developed to address various biological questions, such as sequence identification, functional site characterization and interaction prediction. In this paper, we revisit a collection of 29 state-of-the-art PU learning bioinformatic applications to address various biological questions. Various important aspects are extensively discussed, including PU learning methodology, biological application, classifier design and evaluation strategy. We also comment on the existing issues of PU learning and offer our perspectives for the future development of PU learning applications. We anticipate that our work serves as an instrumental guideline for a better understanding of the PU learning framework in bioinformatics and further developing next-generation PU learning frameworks for critical biological applications.

The pattern of DNA methylation alteration, and its association with the changes of gene expression and alternative splicing during phosphate starvation in tomato.

DNA methylation is changed and associates with gene expression alterations in plant response to phosphate starvation (Pi-), a common stress that impacts plant growth and productivity. However, in the horticultural model species Solanum lycopersicum (tomato), the dynamics of DNA methylation and its relationship with changes in gene transcription and alternative splicing (AS) under Pi- are unknown. Here, we performed integrative methylome and transcriptome analyses of tomato seedlings under Pi-deficient and -sufficient conditions. We found Pi- caused a slight increase in the overall methylation level, with millions of differentially methylated cytosines (DmCs) and a few hundred differentially methylated regions (DMRs). We also identified thousands of differentially expressed (DE) and differential AS (DAS) genes induced by Pi-, and found that DmCs were more abundant in non-expressed genes than in DE or DAS genes. Moreover, DNA methylation alterations weakly correlated with transcription changes but not with DAS events, and hyper-CHH-DMRs overlapping with transposable elements (TEs) were enriched in a subset of Pi starvation response (PSR) genes. We propose that changes in DNA methylation may be associated with the differential expression of some PSR genes, but that most of these changes probably control the expression of nearby TEs, rather than directly affecting the transcription or AS of PSR genes. Besides, the pattern of methylation changes upon Pi- may largely be shaped by TE distributions. Together, our study provides comprehensive insights into the association of DNA methylation with gene transcription and AS under Pi- in tomato and may contribute to unveiling novel roles of epigenetic mechanisms in plant stress response.

The Class III peroxidase gene OsPrx30, transcriptionally modulated by the AT-hook protein OsATH1, mediates rice bacterial blight-induced ROS accumulation.

Class III peroxidases (CIII Prxs) play critical roles in plant immunity by scavenging reactive oxygen species (ROS). However, the functions of CIII Prxs in rice (Oryza sativa L.) immunity are largely unexplored. Here, we report a Prx precursor, OsPrx30, that is responsive to the bacterial blight Xanthomonas oryzae pv. oryzae (Xoo). OsPrx30 was primarily expressed in rice roots, leaves, and stems, and its protein product was mainly localized at the endoplasmic reticulum. Overexpression of OsPrx30 enhanced the plant's susceptibility to Xoo by maintaining a high level of peroxidase (POD) activity and reducing the content of H2 O2 , whereas depletion of OsPrx30 had the opposite effects. Furthermore, we identified an AT-hook transcription factor, OsATH1, that is specifically bound to the OsPrx30 promoter. As observed in plants overexpressing OsPrx30, depletion of OsATH1 enhanced susceptibility to Xoo. Finally, we demonstrated that depletion of OsATH1 increased histone H3 acetylation at the AT-rich region of the OsPrx30 promoter. Taken together, these results reveal a mechanism underlying the POD-induced natural resistance to bacterial diseases and suggest a model for transcription regulation of Prx genes in rice.

NBS-LRR Protein Pik-H4 Interacts with OsBIHD1 to Balance Rice Blast Resistance and Growth by Coordinating Ethylene-Brassinosteroid Pathway.

The regulation of innate immunity and plant growth, along with the trade-off between them, affects the defense and recovery mechanisms of the plant after it is attacked by pathogens. Although it is known that hormonal crosstalk plays a major role in regulating interaction of plant growth and PAMP-triggered immunity, the relationship between plant growth and effector-triggered immunity (ETI) remains unclear. In a large-scale yeast two-hybrid screening for Pik-H4-interacting proteins, a homeodomain transcription factor OsBIHD1 was identified, which is previously known to function in biotic and abiotic stress responses. The knockout of OsBIHD1 in rice lines carrying Pik-H4 largely compromised the resistance of the rice lines to Magnaporthe oryzae, the fungus that causes rice blast. While overexpression of OsBIHD1 resulted in enhanced expression of the pathogenesis-related (PR) and ethylene (ET) synthesis genes. Moreover, OsBIHD1 was also found to directly bind to the promoter region of ethylene-synthesis enzyme OsACO3. In addition, OsBIHD1 overexpression or deficiency provoked dwarfism and reduced brassinosteroid (BR) insensitivity through repressing the expression of several critical genes involved in BR biosynthesis and BR signaling. During M. oryzae infection, transcript levels of the crucial BR catabolic genes (CYP734A2, CYP734A4, and CYP734A6) were significantly up-regulated in OsBIHD1-OX plants. Furthermore, OsBIHD1 was found to be capable of binding to the sequence-specific cis-elements on the promoters of CYP734A2 to suppress the plant growth under fungal invasion. Our results collectively suggest a model that OsBIHD1 is required for Pik-H4-mediated blast resistance through modulating the trade-off between resistance and growth by coordinating brassinosteroid-ethylene pathway.

CONSTANS-Like 9 (OsCOL9) Interacts with Receptor for Activated C-Kinase 1(OsRACK1) to Regulate Blast Resistance through Salicylic Acid and Ethylene Signaling Pathways.

In a previous transcriptome analysis of early response genes in rice during Magnaporthe oryzae infection, we identified a CONSTANS-like (COL) gene OsCOL9. In the present study, we investigated the functional roles of OsCOL9 in blast resistance. OsCOL9 belonged to group II of the COL protein family, and it contained a BB-box and a C-terminal CCT (CONSTANS, COL and TOC1) domain. OsCOL9 was found in the nucleus of rice cells, and it exerted transcriptional activation activities through its middle region (MR). Magnaporthe oryzae infection induced OsCOL9 expression, and transgenic OsCOL9 knock-out rice plants showed increased pathogen susceptibility. OsCOL9 was a critical regulator of pathogen-related genes, especially PR1b, which were also activated by exogenous salicylic acid (SA) and 1-aminocyclopropane-1-carboxylicacid (ACC), the precursor of ethylene (ET). Further analysis indicated that OsCOL9 over-expression increased the expressions of phytohormone biosynthetic genes, NPR1, WRKY45, OsACO1 and OsACS1, which were related to SA and ET biosynthesis. Interestingly, we found that OsCOL9 physically interacted with the scaffold protein OsRACK1 through its CCT domain, and the OsRACK1 expression was induced in response to exogenous SA and ACC as well as M. oryzae infection. Taken together, these results indicated that the COL protein OsCOL9 interacted with OsRACK1, and it enhanced the rice blast resistance through SA and ET signaling pathways.

CONSTANS-like 9 (COL9) delays the flowering time in Oryza sativa by repressing the Ehd1 pathway.

Flowering or heading is one of most important agronomic traits in rice. It has been characterized that CONSTANS (CO) and CONSTANS-like (COL) proteins are critical flowering regulators in response to photoperiodic stress in plants. We have previously identified that the COL family member OsCOL9 can positively enhance the rice blast resistance. In the present study, we aimed to explore the functional role of OsCOL9 in modulating the photoperiodic flowering. Our data showed that overexpression of OsCOL9 delayed the flowering time under both short-day (SD) and long-day (LD) conditions, leading to suppressed expressions of EHd1, RFT and Hd3a at the mRNA Level. OsCOL9 expression exhibited two types of circadian patterns under different daylight conditions, and it could delay the heading date by suppressing the Ehd1 photoperiodic flowering pathway. In contrast, the expressions of previously reported flowering regulators were not significantly changed in OsCOL9 transgenic plants, indicating that OsCOL9 functioned independently of other flowering pathways. In addition, OsCOL9 served as a potential yield gene, and its deficiency reduced the grain number of main panicle in plants. Furthermore, yeast two-hybrid assay indicated that OsCOL9 physically interacted with Receptor for Activated C-kinase 1 (OsRACK1). Rhythmic pattern analysis suggested that OsRACK1 responded to the change of daylight, which was regulated by the circadian clock. Taken together, our results revealed that OsCOL9 could delay the flowering time in rice by repressing the Ehd1 pathway.

4-Coumarate-CoA Ligase-Like Gene OsAAE3 Negatively Mediates the Rice Blast Resistance, Floret Development and Lignin Biosynthesis.

Although adenosine monophosphate (AMP) binding domain is widely distributed in multiple plant species, detailed molecular functions of AMP binding proteins (AMPBPs) in plant development and plant-pathogen interaction remain unclear. In the present study, we identified an AMPBP OsAAE3 from a previous analysis of early responsive genes in rice during Magnaporthe oryzae infection. OsAAE3 is a homolog of Arabidopsis AAE3 in rice, which encodes a 4-coumarate-Co-A ligase (4CL) like protein. A phylogenetic analysis showed that OsAAE3 was most likely 4CL-like 10 in an independent group. OsAAE3 was localized to cytoplasm, and it could be expressed in various tissues. Histochemical staining of transgenic plants carrying OsAAE3 promoter-driven GUS (ß-glucuronidase) reporter gene suggested that OsAAE3 was expressed in all tissues of rice. Furthermore, OsAAE3-OX plants showed increased susceptibility to M. Oryzae, and this finding was attributable to decreased expression of pathogen-related 1a (PR1) and low level of peroxidase (POD) activity. Moreover, OsAAE3 over-expression resulted in increased content of H2O2, leading to programmed cell-death induced by reactive oxygen species (ROS). In addition, OsAAE3 over-expression repressed the floret development, exhibiting dramatically twisted glume and decreased fertility rate of anther. Meanwhile, the expressions of lignin biosynthesis genes were significantly decreased in OsAAE3-OX plants, thereby leading to reduced lignin content. Taken together, OsAAE3 functioned as a negative regulator in rice blast resistance, floret development, and lignin biosynthesis. Our findings further expanded the knowledge in functions of AMBPs in plant floret development and the regulation of rice-fungus interaction.