Results 1 
6 of
6
Learning Parameter Sets for Alignment Advising
"... While the multiple sequence alignment output by an aligner strongly depends on the parameter values used for the alignment scoring function (such as the choice of gap penalties and substitution scores), most users rely on the single default parameter setting provided by the aligner. A different para ..."
Abstract

Cited by 2 (2 self)
 Add to MetaCart
(Show Context)
While the multiple sequence alignment output by an aligner strongly depends on the parameter values used for the alignment scoring function (such as the choice of gap penalties and substitution scores), most users rely on the single default parameter setting provided by the aligner. A different parameter setting, however, might yield a much higherquality alignment for the specific set of input sequences. The problem of picking a good choice of parameter values for specific input sequences is called parameter advising. A parameter advisor has two ingredients: (i) a set of parameter choices to select from, and (ii) an estimator that provides an estimate of the accuracy of the alignment computed by the aligner using a parameter choice. The parameter advisor picks the parameter choice from the set whose resulting alignment has highest estimated accuracy. We consider for the first time the problem of learning the optimal set of parameter choices for a parameter advisor that uses a given accuracy estimator. The optimal set is one that maximizes the expected true accuracy of the resulting parameter advisor, averaged over a collection of training data. While we prove that learning an optimal set for an advisor is NPcomplete, we show there is a natural approximation algorithm for this problem, and prove a tight bound on its approximation ratio. Experiments with an implementation of this approximation algorithm on biological benchmarks, using various accuracy estimators from the literature, show it finds sets for advisors that are surprisingly close to optimal. Furthermore, the resulting parameter advisors are significantly more accurate in practice than simply aligning with a single default parameter choice.
Ensemble Multiple Sequence Alignment via Advising
"... The multiple sequence alignments computed by an aligner for different settings of its parameters, as well as the alignments computed by different aligners using their default settings, can differ markedly in accuracy. Parameter advising is the task of choosing a parameter setting for an aligner to ..."
Abstract
 Add to MetaCart
The multiple sequence alignments computed by an aligner for different settings of its parameters, as well as the alignments computed by different aligners using their default settings, can differ markedly in accuracy. Parameter advising is the task of choosing a parameter setting for an aligner to maximize the accuracy of the resulting alignment. We extend parameter advising to aligner advising, which in contrast chooses among a set of aligners to maximize accuracy. In the context of aligner advising, default advising selects from a set of aligners that are using their default settings, while general advising selects both the aligner and its parameter setting. In this paper, we apply aligner advising for the first time, to create a true ensemble aligner. Through crossvalidation experiments on benchmark protein sequence alignments, we show that parameter advising boosts an aligner’s accuracy beyond its default setting for virtually all of the standard aligners currently used in practice. Furthermore, aligner advising with a collection of aligners further improves upon parameter advising with any single aligner, though surprisingly the performance of default advising on testing data is actually superior to general advising due to less overfitting to training data. The new ensemble aligner that results from aligner advising is significantly more accurate than the best single default aligner, especially on hardtoalign sequences. This successfully demonstrates how to construct out of a collection of individual aligners, a more accurate ensemble aligner.
IEEE/ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS 1 Learning ParameterAdvising Sets for Multiple Sequence Alignment
"... Abstract—While the multiple sequence alignment output by an aligner strongly depends on the parameter values used for the alignment scoring function (such as the choice of gap penalties and substitution scores), most users rely on the single default parameter setting provided by the aligner. A diffe ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract—While the multiple sequence alignment output by an aligner strongly depends on the parameter values used for the alignment scoring function (such as the choice of gap penalties and substitution scores), most users rely on the single default parameter setting provided by the aligner. A different parameter setting, however, might yield a much higherquality alignment for the specific set of input sequences. The problem of picking a good choice of parameter values for specific input sequences is called parameter advising. A parameter advisor has two ingredients: (i) a set of parameter choices to select from, and (ii) an estimator that provides an estimate of the accuracy of the alignment computed by the aligner using a parameter choice. The parameter advisor picks the parameter choice from the set whose resulting alignment has highest estimated accuracy. We consider for the first time the problem of learning the optimal set of parameter choices for a parameter advisor that uses a given accuracy estimator. The optimal set is one that maximizes the expected true accuracy of the resulting parameter advisor, averaged over a collection of training data. While we prove that learning an optimal set for an advisor is NPcomplete, we show there is a natural approximation algorithm for this problem, and prove a tight bound on its approximation ratio. Experiments with an implementation of this approximation algorithm on biological benchmarks, using various accuracy estimators from the literature, show it finds sets for advisors that are surprisingly close to optimal. Furthermore, the resulting parameter advisors are significantly more accurate in practice than simply aligning with a single default parameter choice. Index Terms—Multiple sequence alignment, alignment scoring functions, parameter values, accuracy estimation, parameter advising. F 1
Research Articles Accuracy Estimation and Parameter Advising for Protein Multiple Sequence Alignment
"... We develop a novel and general approach to estimating the accuracy of multiple sequence alignments without knowledge of a reference alignment, and use our approach to address a new task that we call parameter advising: the problem of choosing values for alignment scoring function parameters from a g ..."
Abstract
 Add to MetaCart
We develop a novel and general approach to estimating the accuracy of multiple sequence alignments without knowledge of a reference alignment, and use our approach to address a new task that we call parameter advising: the problem of choosing values for alignment scoring function parameters from a given set of choices to maximize the accuracy of a computed alignment. For protein alignments, we consider twelve independent features that contribute to a quality alignment. An accuracy estimator is learned that is a polynomial function of these features; its coefficients are determined by minimizing its error with respect to true accuracy using mathematical optimization. Compared to prior approaches for estimating accuracy, our new approach (a) introduces novel feature functions that measure nonlocal properties of an alignment yet are fast to evaluate, (b) considers more general classes of estimators beyond linear combinations of features, and (c) develops new regression formulations for learning an estimator from examples; in addition, for parameter advising, we (d) determine the optimal parameter set of a given cardinality, which specifies the best parameter values from which to choose. Our estimator, which we call Facet (for ‘‘featurebased accuracy estimator’’), yields a parameter advisor that on the hardest benchmarks provides more than a 27 % improvement in accuracy over the best default parameter choice, and for parameter advising significantly outperforms the best prior approaches to assessing alignment quality. Key words: sequence alignment, accuracy assessment, parameter choice, machine learning, feature functions.
METHODOLOGY ARTICLE Open Access
"... Efficient representation of uncertainty in multiple sequence alignments using directed acyclic graphs Joseph L Herman1,2*, Ádám Novák1, Rune Lyngsø1, Adrienn Szabó3, István Miklós3,4 and Jotun Hein1 Background: A standard procedure in many areas of bioinformatics is to use a single multiple sequence ..."
Abstract
 Add to MetaCart
(Show Context)
Efficient representation of uncertainty in multiple sequence alignments using directed acyclic graphs Joseph L Herman1,2*, Ádám Novák1, Rune Lyngsø1, Adrienn Szabó3, István Miklós3,4 and Jotun Hein1 Background: A standard procedure in many areas of bioinformatics is to use a single multiple sequence alignment (MSA) as the basis for various types of analysis. However, downstream results may be highly sensitive to the alignment used, and neglecting the uncertainty in the alignment can lead to significant bias in the resulting inference. In recent years, a number of approaches have been developed for probabilistic sampling of alignments, rather than simply generating a single optimum. However, this type of probabilistic information is currently not widely used in the context of downstream inference, since most existing algorithms are set up to make use of a single alignment. Results: In this work we present a framework for representing a set of sampled alignments as a directed acyclic graph (DAG) whose nodes are alignment columns; each path through this DAG then represents a valid alignment. Since the probabilities of individual columns can be estimated from empirical frequencies, this approach enables samplebased estimation of posterior alignment probabilities. Moreover, due to conditional independencies between