Unsupervised Analysis Technique Research Articles

Phenotypic Diversity of Pomegranate Cultivars: Discriminating Power of Some Morphological and Fruit Chemical Characteristics

In modern agricultural production, where a small number of commercial cultivars dominate, the collection, evaluation, and preservation of germplasm are important tasks to reduce the erosion of genes and preserve biodiversity. The aim of this study is to characterize the morphological and fruit chemical properties of the pomegranate germplasm grown on the East Adriatic coast, including the commercial cultivars ‘Hicaznar’, ‘Granada’, and ‘Wonderful’, and to highlight the characteristics with the greatest discriminating power. The characterization of the tree, leaf, flower, arils, seed, and juice was carried out using the UPOV descriptor. The colors of the peel, arils, and juice were analyzed according to the CIEL*a*b* method, total soluble solids were measured using refractometers, and total acidity was determined by titration with 0.1 M NaOH. The research results showed significant diversity between the cultivars, which were grouped into several clusters using an unsupervised analysis technique. Factors such as plant vigor, plant growth habit, predominant number of leaves per node on young shoots, crown type, fruit shape, fruit shape in cross-section, peel weight, total aril weight, aril weight, number of arils per fruit, seed length and width, seed yield, total acidity, TSS/TA ratio, and color parameters of the peel, arils, and juice showed high variability, indicating their strong discriminating power in determining the phenotypic diversity of pomegranate.

Biomedical Document Clustering Based on Accelerated Symbiotic Organisms Search Algorithm

Clustering is an important unsupervised analysis technique for big data mining. It finds its application in several domains including biomedical documents of the MEDLINE database. Document clustering algorithms based on metaheuristics is an active research area. However, these algorithms suffer from the problems of getting trapped in local optima, need many parameters to adjust, and the documents should be indexed by a high dimensionality matrix using the traditional vector space model. In order to overcome these limitations, in this paper a new documents clustering algorithm (ASOS-LSI) with no parameters is proposed. It is based on the recent symbiotic organisms search metaheuristic (SOS) and enhanced by an acceleration technique. Furthermore, the documents are represented by semantic indexing based on the famous latent semantic indexing (LSI). Conducted experiments on well-known biomedical documents datasets show the significant superiority of ASOS-LSI over five famous algorithms in terms of compactness, f-measure, purity, misclassified documents, entropy, and runtime.

Application of unsupervised analysis techniques to lung cancer patient data.

This study applies unsupervised machine learning techniques for classification and clustering to a collection of descriptive variables from 10,442 lung cancer patient records in the Surveillance, Epidemiology, and End Results (SEER) program database. The goal is to automatically classify lung cancer patients into groups based on clinically measurable disease-specific variables in order to estimate survival. Variables selected as inputs for machine learning include Number of Primaries, Age, Grade, Tumor Size, Stage, and TNM, which are numeric or can readily be converted to numeric type. Minimal up-front processing of the data enables exploring the out-of-the-box capabilities of established unsupervised learning techniques, with little human intervention through the entire process. The output of the techniques is used to predict survival time, with the efficacy of the prediction representing a proxy for the usefulness of the classification. A basic single variable linear regression against each unsupervised output is applied, and the associated Root Mean Squared Error (RMSE) value is calculated as a metric to compare between the outputs. The results show that self-ordering maps exhibit the best performance, while k-Means performs the best of the simpler classification techniques. Predicting against the full data set, it is found that their respective RMSE values (15.591 for self-ordering maps and 16.193 for k-Means) are comparable to supervised regression techniques, such as Gradient Boosting Machine (RMSE of 15.048). We conclude that unsupervised data analysis techniques may be of use to classify patients by defining the classes as effective proxies for survival prediction.

Abstract 780: Multi-omic profiling of gliomas reveals distinct DNA methylation changes at tumor recurrence

Abstract Varying possibilities of tumor relapse for lower grade glioma (LGG, WHO grade II and III) and glioblastoma (GBM, WHO grade IV) have led to heterogenous clinical outcomes for patients. Our current study aims to establish a molecular profile of recurrence of matched primary and recurrent LGG (n = 28) and recurrent GBM (n = 24) tumor samples. The Cancer Genome Atlas (TCGA) has comprehensively profiled these matched primary/recurrent tumor sets; whole genomes, coding exomes, methylomes, and transcriptomes have been sequenced, and the samples have undergone targeted profiling of the proteome. While unsupervised analysis techniques have not led to a clear recurrence signature, supervised analysis methods have revealed interesting patterns. Protein profiling has shown that recurrent gliomas retain the overall molecular signature of their primary counterpart, but the DNA damage response, apoptosis and RTK pathways are downregulated in the recurrent gliomas, in contrast to RAS/MAPK, PI3K/AKT, and EMT pathways, which are upregulated. Whole genome sequencing and rearrangement analysis have revealed increased genomic complexity among most recurrent gliomas as well as new fusions of interest in recurrent LGG samples (PTPRZ1-MET and involving ATRX). Using genome-wide Illumina HumanMethylation 450K data we observed that 78.6% of LGGs showed depletion of DNA methylation at recurrence and 50% of GBM tumors showed an enrichment of DNA methylation at recurrence. Patient centric enrichment analysis allowed us to discover a candidate biological subgroup characterized by a subset of LGG recurrences (50%) exhibiting an aberrant CpG methylation loss at inintergenic opensea regions when compared with canonical CpG islands and shores (fold > 1.3 and confidence intervals of 99%). Importantly, inspection of CpG sites significantly hypomethylated at openseas showed that this pronounced epigenetic signature maps to candidate TSS distal and hypomethylated enhancers. The gene-targets of these hypomethylated CpG sites show a corresponding up-regulation of expression. Pathway analysis has demonstrated that these upregulated genes are involved in cellular growth and proliferation, cellular function and maintenance, and cell cycle regulation. Our results provide evidence that DNA methylation may represent a stable signature of glioma recurrence and that the crosstalk between DNA hypomethylation at openseas and chromosomal instability may be involved in glioma recurrence. We plan to further integrate our findings between data types and correlate with treatment and patient clinical outcome. Citation Format: Lindsay C. Stetson, Camila Ferreira de Souza, Tathiane Maistro Malta, Thais Sarraf Sabedot, Quinn Ostrom, Peter Liao, Daniela Pretti da Cunha Tirapelli, Luciano Neder, Carlos Gilberto Carlotti, Rehan Akbani, Sofie Salama, Laila Poisson, Daniel Brat, The Cancer Genome Atlas Network, Houtan Noushmehr, Jill Barnholtz-Sloan. Multi-omic profiling of gliomas reveals distinct DNA methylation changes at tumor recurrence. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 780.

A Scale Independent Selection Process for 3D Object Recognition in Cluttered Scenes

During the last years a wide range of algorithms and devices have been made available to easily acquire range images. The increasing abundance of depth data boosts the need for reliable and unsupervised analysis techniques, spanning from part registration to automated segmentation. In this context, we focus on the recognition of known objects in cluttered and incomplete 3D scans. Locating and fitting a model to a scene are very important tasks in many scenarios such as industrial inspection, scene understanding, medical imaging and even gaming. For this reason, these problems have been addressed extensively in the literature. Several of the proposed methods adopt local descriptor-based approaches, while a number of hurdles still hinder the use of global techniques. In this paper we offer a different perspective on the topic: We adopt an evolutionary selection algorithm that seeks global agreement among surface points, while operating at a local level. The approach effectively extends the scope of local descriptors by actively selecting correspondences that satisfy global consistency constraints, allowing us to attack a more challenging scenario where model and scene have different, unknown scales. This leads to a novel and very effective pipeline for 3D object recognition, which is validated with an extensive set of experiments and comparisons with recent techniques at the state of the art.