Proteome Analyst: custom predictions with explanations in a web-based tool for high-throughput proteome annotations

Abstract

Proteome Analyst (PA) (http://www.cs.ualberta.ca/~bioinfo/PA/) is a publicly available, high-throughput, web-based system for predicting various properties of each protein in an entire proteome. Using machine-learned classifiers, PA can predict, for example, the GeneQuiz general function and Gene Ontology (GO) molecular function of a protein. In addition, PA is currently the most accurate and most comprehensive system for predicting subcellular localization, the location within a cell where a protein performs its main function. Two other capabilities of PA are notable. First, PA can create a custom classifier to predict a new property, without requiring any programming, based on labeled training data (i.e. a set of examples, each with the correct classification label) provided by a user. PA has been used to create custom classifiers for potassium-ion channel proteins and other general function ontologies. Second, PA provides a sophisticated explanation feature that shows why one prediction is chosen over another. The PA system produces a Naïve Bayes classifier, which is amenable to a graphical and interactive approach to explanations for its predictions; transparent predictions increase the user's confidence in, and understanding of, PA.

Computational analysis of the human HSPH/HSPA/DNAJ family and cloning of a human HSPH/HSPA/DNAJ expression library

“…Subcellular localization analysis Predictions on subcellular localizations were performed using pSort, pTarget, CELLO, Multiloc, and Proteome analyst (Szafron et al 2004;Yu et al 2006;Guda 2006;Hoglund et al 2006;Horton et al 2007). Sequences from complete gene families were uploaded as fasta files.…”

“…We selected several publicly available localization prediction methods, which accept large batches of protein sequences and which were able to predict all of the major subcellular localizations. The selected methods were Wolf PSORT (Horton et al 2007), pTarget (Guda 2006), CELLO (Yu et al 2006), Multiloc (Hoglund et al 2006) and Proteome Analyst (Szafron et al 2004). In addition, we searched the human protein database (Mishra et al 2006) for experimentally verified HSP localizations.…”

A novel method for predicting protein subcellular localization based on pseudo amino acid composition

“…In this work, the ELM-PCA predictor is compared with other classification methods such as Randomtree (21), J48 (22), Naivebayes (23) and five popular web-servers PSLpred (24), PSORTb (25), Cello (26), Proteome Analyst (PA) (27) and SOSUI (28). Also, we construct another independent dataset containing 601 proteins, of which 140 are of the cytoplasm, 74 of the extracellular locations, 280 of the inner membrane, 38 of the outer membrane, and 69 of the periplasm, to assess effectiveness of our method.…”

Bayesian methods for proteomics

“…A web-based program (http://www.cs.ualberta.ca/,bioinfo/PA/), PA predicts characteristics of proteins across a proteome [49]. Its authors report that it is currently the most accurate program in predicting subcellular localization.…”

“…Using a Bayesian framework, however, it boasts two additional features: the ability to create user-defined classifications without programming as well as providing an intuitive explanation as to why one hypothesis was picked over another (see Fig. 8) [49].…”