Abstract

Abstract

Abstract

Abstract
This paper reports new results in adopting entropy concepts to the training of neural networks to perform wind power prediction as a function of wind characteristics (speed and direction) in wind parks connected to a power grid. Renyi's entropy is combined with a Parzen windows estimation of the error pdf to form the basis of two criteria (minimum entropy and maximum correntropy) under which neural networks are trained. The results are favorably compared in online and offline training with the traditional minimum square error (MSE) criterion. Real case examples for two distinct wind parks are presented.

Abstract
Background: Comparative studies using hundreds of sequences can give a detailed picture of the evolution of a given gene family. Nevertheless, retrieving only the sequences of interest from public databases can be difficult, in particular, when working with highly divergent sequences. The difficulty increases substantially when one wants to include in the study sequences from many (or less well studied) species whose genomes are non-annotated or incompletely annotated. Methodology/Principal Findings: In this work we evaluate the usefulness of different approaches of gene retrieval and classification, using the distal-less (DLX) gene family as a test case. Furthermore, we evaluate whether the use of a large number of gene sequences from a wide range of animal species, the use of multiple alternative alignments, and the use of amino acids aligned with high confidence only, is enough to recover the accepted DLX evolutionary history. Conclusions/Significance: The canonical DLX homeobox gene sequence here derived, together with the characteristic amino acid variants here identified in the DLX homeodomain region, can be used to retrieve and classify DLX genes in a simple and efficient way. A program is made available that allows the easy retrieval of synteny information that can be used to classify gene sequences. Maximum likelihood trees using hundreds of sequences can be used for gene identification. Nevertheless, for the DLX case, the proposed DLX evolutionary is not recovered even when multiple alignment algorithms are used.

Abstract
Multiple independent recruitments of the S-pollen component (always an F-box gene) during RNase-based gametophytic self-incompatibility evolution have recently been suggested. Therefore, different mechanisms could be used to achieve the rejection of incompatible pollen in different plant families. This hypothesis is, however, mainly based on the interpretation of phylogenetic analyses, using a small number of divergent nucleotide sequences. In this work we show, based on a large collection of F-box S-like sequences, that the inferred relationship of F-box S-pollen and F-box S-like sequences is dependent on the sequence alignment software and phylogenetic method used. Thus, at present, it is not possible to address the phylogenetic relationship of F-box S-pollen and S-like sequences from different plant families. In Petunia and Malus/ Pyrus the putative S-pollen gene(s) show(s) variability patterns different than expected for an S-pollen gene, raising the question of false identification. Here we show that in Petunia, the unexpected features of the putative S-pollen gene are not incompatible with this gene's being the S-pollen gene. On the other hand, it is very unlikely that the Pyrus SFBB-gamma gene is involved in specificity determination.