Systematic feature evaluation for gene name recognition

Abstract

In task 1A of the BioCreAtIvE evaluation, systems had to be devised that recognize words and phrases forming gene or protein names in natural language sentences. We approach this problem by building a word classification system based on a sliding window approach with a Support Vector Machine, combined with a pattern-based post-processing for the recognition of phrases. The performance of such a system crucially depends on the type of features chosen for consideration by the classification method, such as pre-or postfixes, character n-grams, patterns of capitalization, or classification of preceding or following words. We present a systematic approach to evaluate the performance of different feature sets based on recursive feature elimination, RFE. Based on a systematic reduction of the number of features used by the system, we can quantify the impact of different feature sets on the results of the word classification problem. This helps us to identify descriptive features, to learn about the structure of the problem, and to design systems that are faster and easier to understand. We observe that the SVM is robust to redundant features. RFE improves the performance by 0.7%, compared to using the complete set of attributes. Moreover, a performance that is only 2.3% below this maximum can be obtained using fewer than 5% of the features. Background and overview

Genomic Characterization of Variable Surface Antigens Reveals a Telomere Position Effect as a Prerequisite for RNA Interference-Mediated Silencing in Paramecium tetraurelia

“…Transmembrane domains were predicted by TMHMM as described in reference 65. Graphic alignments were calculated by GATA using default settings (66). Amino acid sequences of SAgs were aligned with ClustalW, and the neighbor joining tree was calculated with MEGA4 (67) with 1,000 bootstrap replicates, based on p distances and after pairwise deletion of gaps.…”

An Overview of Bioinformatics Tools and Resources in Allergy

“…An extensive number of customized kernels (functions that quantify the similarity between sequences) have been proposed and are more effective than general-purpose kernels (e.g., polynomial and radial basis function kernels) [11]. Classifiers developed to explicitly optimize the area under ROC curve (AUC) have been shown to be more useful than to optimize the classification error when the AUC is the performance metric of interest [12], and ensemble learning methods (e.g., boosting and bagging) have been shown effective in improving the performance of a single learning algorithm [13]. Now, with the aid of bioinformatics and machine learning, we are able to narrow protein of our interest, predict epitope using computation, increase allergy diagnostic accuracy, and drastically reduce the number of wet lab experiments.…”

Optimizing syntax patterns for discovering protein-protein interactions

“…Essentially, we use a token-based SVM classifier with expansion to noun-phrases to identify names referring to proteins [5]. This tagger was trained on the BioCreAtIvE Task 1A data [1,15].…”