Search

The increasingly larger quantities of information generated in the world over the last few years, has led to the emergence of the paradigm known as Big Data. The analysis of those vast quantities of data has become an important task in science and business in order to turn that information into a valuable asset. Many data analysis tasks involves the use of machine learning techniques during the model creation step and the goal of these predictive models consists on achieving the highest possible accuracy to predict new samples, and for this reason there is high interest in selecting the most suitable algorithm for a specific dataset. This trend is known as model selection and it has been widely studied in datasets of common size, but poorly explored in the Big Data context. As an effort to explore in this direction this work propose an algorithm for model selection in Big Data.

There is a natural protein form, insoluble and resistant to proteolysis, adopted by many proteins independently of their amino acid sequences via specific misfolding-aggregation process. This dynamic process occurs in parallel with or as an alternative to physiologic folding, generating toxic protein aggregates that are deposited and accumulated in various organs and tissues. These proteinaceous deposits typically represent bundles of β-sheet-enriched fibrillar species known as the amyloid fibrils that are responsible for serious pathological conditions, including but not limited to neurodegenerative diseases, grouped under the term amyloidoses. The proteins that might adopt this fibrillar conformation are some globular proteins and natively unfolded (or intrinsically disordered) proteins. Our work shows that intrinsically disordered and intrinsically ordered proteins can be reliably identified, discriminated, and differentiated by analyzing their polarity profiles generated using a computational tool known as the polarity index method (Polanco & Samaniego, 2009; Polanco et al., 2012; 2013; 2013a; 2014; 2014a; 2014b; 2014c; 2014d). We also show that proteins expressed in neurons can be differentiated from proteins in these two groups based on their polarity profiles, and also that this computational tool can be used to identify proteins associated with amyloidoses. The efficiency of the proposed method is high (i.e. 70%) as evidenced by the analysis of peptides and proteins in the APD2 database (2012), AVPpred database (2013), and CPPsite database (2013), the set of selective antibacterial peptides from del Rio et al. (2001), the sets of natively unfolded and natively folded proteins from Oldfield et al. (2005), the set of human revised proteins expressed in neurons, and non-human revised proteins expressed in neurons, from the Uniprot database (2014), and also the set of amyloidogenic proteins from the AmyPDB database (2014).

In this paper we present the morphological operator pecstrum, or pattern spectrum, as a feature extractor of discriminating characteristics in microscopic leukocytes images for classification purposes. Pecstrum provides an excellent quantitative analysis to model the morphological evolution of nuclei in blood white cells, or leukocytes. According to their maturity stage, leukocytes have been classified by medical experts in six categories, from myeloblast to polymorphonuclear corresponding to the youngest and oldest extremes, respectively. A feature vector based on the pattern spectrum, normalized area, and nucleus - cytoplasm area ratio, was tested using a multilayer perceptron neural network trained by backpropagation, and a Support Vector Machine algorithm. Results from Euclidean distance and k-nearest neighbor classifiers are also reported as reference for comparison purposes. A recognition rate of 87% was obtained in the best case, using 36 patterns for training and 18 for testing, with a three-fold validation scheme. Additional experiments exploring larger databases are currently in progress.

In this paper we present preliminary results obtained from the application of morphological operator pecstrum, for the extraction of discriminating characteristics in leukocytes and similar artificial images. Experts have identified six categories of leukocytes, very similar in shape and size, which makes them extremely difficult to distinguish automatically or even by non-expert humans. A feature vector based on a 7-component pecstrum, normalized area, and nucleus-cytoplasm area ratio, was tested using 4 kinds of recognizers: Euclidean distance, k-nearest neighbor, back propagation neural net and support vector machine. Using 36 patterns for training and 18 for testing, recognition of 87% was obtained in the best case, which is encouraging, given the complexity of the problem. The amount of samples used at this point for experiments is not statistically representative, however these results are promising and more experiments will be carried out.