Tensor LRR and sparse coding-based subspace clustering

Subspace clustering groups a set of samples from a union of several linear subspaces into clusters, so that the samples in the same cluster are drawn from the same linear subspace. In the majority of the existing work on subspace clustering, clusters are built based on feature information, while sample correlations in their original spatial structure are simply ignored. Besides, original high-dimensional feature vector contains noisy/redundant information, and the time complexity grows exponentially with the number of dimensions. To address these issues, we propose a tensor low-rank representation (TLRR) and sparse coding-based (TLRRSC) subspace clustering method by simultaneously considering feature information and spatial structures. TLRR seeks the lowest rank representation over original spatial structures along all spatial directions. Sparse coding learns a dictionary along feature spaces, so that each sample can be represented by a few atoms of the learned dictionary. The affinity matrix used for spectral clustering is built from the joint similarities in both spatial and feature spaces. TLRRSC can well capture the global structure and inherent feature information of data, and provide a robust subspace segmentation from corrupted data. Experimental results on both synthetic and real-world data sets show that TLRRSC outperforms several established state-of-the-art methods.

The pgip family in soybean and three other legume species: evidence for a birth-and-death model of evolution

Background Polygalacturonase-inhibiting proteins (PGIPs) are leucine-rich repeat (LRR) plant cell wall glycoproteins involved in plant immunity. They are typically encoded by gene families with a small number of gene copies whose evolutionary origin has been poorly investigated. Here we report the complete characterization of the full complement of the pgip family in soybean (Glycine max [L.] Merr.) and the characterization of the genomic region surrounding the pgip family in four legume species. Results BAC clone and genome sequence analyses showed that the soybean genome contains two pgip loci. Each locus is composed of three clustered genes that are induced following infection with the fungal pathogen Sclerotinia sclerotiorum (Lib.) de Bary, and remnant sequences of pgip genes. The analyzed homeologous soybean genomic regions (about 126 Kb) that include the pgip loci are strongly conserved and this conservation extends also to the genomes of the legume species Phaseolus vulgaris L., Medicago truncatula Gaertn. and Cicer arietinum L., each containing a single pgip locus. Maximum likelihood-based gene trees suggest that the genes within the pgip clusters have independently undergone tandem duplication in each species. Conclusions The paleopolyploid soybean genome contains two pgip loci comprised in large and highly conserved duplicated regions, which are also conserved in bean, M. truncatula and C. arietinum. The genomic features of these legume pgip families suggest that the forces driving the evolution of pgip genes follow the birth-and-death model, similar to that proposed for the evolution of resistance (R) genes of NBS-LRR-type.

Global transcriptome analysis and identification of differentially expressed genes after infection of Fragaria vesca with powdery mildew (Podosphaera aphanis)

Background: Podosphaera aphanis, the causal agent of strawberry powdery mildew causes significant economic loss worldwide. Methods: We used the diploid strawberry species Fragaria vesca as a model to study plant pathogen interactions. RNA-seq was employed to generate a transcriptome dataset from two accessions, F. vesca ssp. vesca Hawaii 4 (HW) and F. vesca f. semperflorens Yellow Wonder 5AF7 (YW) at 1 d (1 DAI) and 8 d (8 DAI) after infection. Results: Of the total reads identified about 999 million (92%) mapped to the F. vesca genome. These transcripts were derived from a total of 23,470 and 23,464 genes in HW and YW, respectively from the three time points (control, 1 and 8 DAI). Analysis identified 1,567, 1,846 and 1,145 up-regulated genes between control and 1 DAI, control and 8 DAI, and 1 and 8 DAI, respectively in HW. Similarly, 1,336, 1,619 and 968 genes were up-regulated in YW. Also 646, 1,098 and 624 down-regulated genes were identified in HW, while 571, 754 and 627 genes were down-regulated in YW between all three time points, respectively. Conclusion: Investigation of differentially expressed genes (log2 fold changes �5) between control and 1 DAI in both HW and YW identified a large number of genes related to secondary metabolism, signal transduction; transcriptional regulation and disease resistance were highly expressed. These included flavonoid 3´-monooxygenase, peroxidase 15, glucan endo-1,3-β-glucosidase 2, receptor-like kinases, transcription factors, germin-like proteins, F-box proteins, NB-ARC and NBS-LRR proteins. This is the first application of RNA-seq to any pathogen interaction in strawberry

Low rank representation on Riemannian manifold of symmetric positive definite matrices

Sparse coding aims to find a more compact representation based on a set of dictionary atoms. A well-known technique looking at 2D sparsity is the low rank representation (LRR). However, in many computer vision applications, data often originate from a manifold, which is equipped with some Riemannian geometry. In this case, the existing LRR becomes inappropriate for modeling and incorporating the intrinsic geometry of the manifold that is potentially important and critical to applications. In this paper, we generalize the LRR over the Euclidean space to the LRR model over a specific Rimannian manifold—the manifold of symmetric positive matrices (SPD). Experiments on several computer vision datasets showcase its noise robustness and superior performance on classification and segmentation compared with state-of-the-art approaches.