Remote homology detection based on oligomer distances

A Comprehensive Review on Machine Learning Techniques for Protein Family Prediction

Proteomics is a field dedicated to the analysis of proteins in cells, tissues, and organisms, aiming to gain insights into their structures, functions, and interactions. A crucial aspect within proteomics is protein family prediction, which involves identifying evolutionary relationships between proteins by examining similarities in their sequences or structures. This approach holds great potential for applications such as drug discovery and functional annotation of genomes. However, current methods for protein family prediction have certain limitations, including limited accuracy, high false positive rates, and challenges in handling large datasets. Some methods also rely on homologous sequences or protein structures, which introduce biases and restrict their applicability to specific protein families or structures. To overcome these limitations, researchers have turned to machine learning (ML) approaches that can identify connections between protein features and simplify complex high-dimensional datasets. This paper presents a comprehensive survey of articles that employ various ML techniques for predicting protein families. The primary objective is to explore and improve ML techniques specifically for protein family prediction, thus advancing future research in the field. Through qualitative and quantitative analyses of ML techniques, it is evident that multiple methods utilizing a range of classifiers have been applied for protein family prediction. However, there has been limited focus on developing novel classifiers for protein family classification, highlighting the urgent need for improved approaches in this area. By addressing these challenges, this research aims to enhance the accuracy and effectiveness of protein family prediction, ultimately facilitating advancements in proteomics and its diverse applications.

EnsembleFam: towards more accurate protein family prediction in the twilight zone

Background

Current protein family modeling methods like profile Hidden Markov Model (pHMM), k-mer based methods, and deep learning-based methods do not provide very accurate protein function prediction for proteins in the twilight zone, due to low sequence similarity to reference proteins with known functions.

Results

We present a novel method EnsembleFam, aiming at better function prediction for proteins in the twilight zone. EnsembleFam extracts the core characteristics of a protein family using similarity and dissimilarity features calculated from sequence homology relations. EnsembleFam trains three separate Support Vector Machine (SVM) classifiers for each family using these features, and an ensemble prediction is made to classify novel proteins into these families. Extensive experiments are conducted using the Clusters of Orthologous Groups (COG) dataset and G Protein-Coupled Receptor (GPCR) dataset. EnsembleFam not only outperforms state-of-the-art methods on the overall dataset but also provides a much more accurate prediction for twilight zone proteins.

Conclusions

EnsembleFam, a machine learning method to model protein families, can be used to better identify members with very low sequence homology. Using EnsembleFam protein functions can be predicted  using just sequence information with better accuracy than state-of-the-art methods.

Graph Theory-Based Sequence Descriptors as Remote Homology Predictors

Alignment-free (AF) methodologies have increased in popularity in the last decades as alternative tools to alignment-based (AB) algorithms for performing comparative sequence analyses. They have been especially useful to detect remote homologs within the twilight zone of highly diverse gene/protein families and superfamilies. The most popular alignment-free methodologies, as well as their applications to classification problems, have been described in previous reviews. Despite a new set of graph theory-derived sequence/structural descriptors that have been gaining relevance in the detection of remote homology, they have been omitted as AF predictors when the topic is addressed. Here, we first go over the most popular AF approaches used for detecting homology signals within the twilight zone and then bring out the state-of-the-art tools encoding graph theory-derived sequence/structure descriptors and their success for identifying remote homologs. We also highlight the tendency of integrating AF features/measures with the AB ones, either into the same prediction model or by assembling the predictions from different algorithms using voting/weighting strategies, for improving the detection of remote signals. Lastly, we briefly discuss the efforts made to scale up AB and AF features/measures for the comparison of multiple genomes and proteomes. Alongside the achieved experiences in remote homology detection by both the most popular AF tools and other less known ones, we provide our own using the graphical-numerical methodologies, MARCH-INSIDE, TI2BioP, and ProtDCal. We also present a new Python-based tool (SeqDivA) with a friendly graphical user interface (GUI) for delimiting the twilight zone by using several similar criteria.

DeepFam: deep learning based alignment-free method for protein family modeling and prediction

Motivation

A large number of newly sequenced proteins are generated by the next-generation sequencing technologies and the biochemical function assignment of the proteins is an important task. However, biological experiments are too expensive to characterize such a large number of protein sequences, thus protein function prediction is primarily done by computational modeling methods, such as profile Hidden Markov Model (pHMM) and k-mer based methods. Nevertheless, existing methods have some limitations; k-mer based methods are not accurate enough to assign protein functions and pHMM is not fast enough to handle large number of protein sequences from numerous genome projects. Therefore, a more accurate and faster protein function prediction method is needed.

Results

In this paper, we introduce DeepFam, an alignment-free method that can extract functional information directly from sequences without the need of multiple sequence alignments. In extensive experiments using the Clusters of Orthologous Groups (COGs) and G protein-coupled receptor (GPCR) dataset, DeepFam achieved better performance in terms of accuracy and runtime for predicting functions of proteins compared to the state-of-the-art methods, both alignment-free and alignment-based methods. Additionally, we showed that DeepFam has a power of capturing conserved regions to model protein families. In fact, DeepFam was able to detect conserved regions documented in the Prosite database while predicting functions of proteins. Our deep learning method will be useful in characterizing functions of the ever increasing protein sequences.

Availability and implementation

Codes are available at https://bhi-kimlab.github.io/DeepFam.

ProtDet-CCH: Protein Remote Homology Detection by Combining Long Short-Term Memory and Ranking Methods

As one of the most challenging tasks in sequence analysis, protein remote homology detection has been extensively studied. Methods based on discriminative models and ranking approaches have achieved the state-of-the-art performance, and these two kinds of methods are complementary. In this study, three LSTM models have been applied to construct the predictors for protein remote homology detection, including ULSTM, BLSTM, and CNN-BLSTM. They are able to automatically extract the local and global sequence order information. Combined with PSSMs, the CNN-BLSTM achieved the best performance among the three LSTM-based models. We named this method as CNN-BLSTM-PSSM. Finally, a new method called ProtDet-CCH was proposed by combining CNN-BLSTM-PSSM and a ranking method HHblits. Tested on a widely used SCOP benchmark dataset, ProtDet-CCH achieved an ROC score of 0.998, and an ROC50 score of 0.982, significantly outperforming other existing state-of-the-art methods. Experimental results on two updated SCOPe independent datasets showed that ProtDet-CCH can achieve stable performance. Furthermore, our method can provide useful insights for studying the features and motifs of protein families and superfamilies. It is anticipated that ProtDet-CCH will become a very useful tool for protein remote homology detection.

Alignment-free method for DNA sequence clustering using Fuzzy integral similarity

A larger amount of sequence data in private and public databases produced by next-generation sequencing put new challenges due to limitation associated with the alignment-based method for sequence comparison. So, there is a high need for faster sequence analysis algorithms. In this study, we developed an alignment-free algorithm for faster sequence analysis. The novelty of our approach is the inclusion of fuzzy integral with Markov chain for sequence analysis in the alignment-free model. The method estimate the parameters of a Markov chain by considering the frequencies of occurrence of all possible nucleotide pairs from each DNA sequence. These estimated Markov chain parameters were used to calculate similarity among all pairwise combinations of DNA sequences based on a fuzzy integral algorithm. This matrix is used as an input for the neighbor program in the PHYLIP package for phylogenetic tree construction. Our method was tested on eight benchmark datasets and on in-house generated datasets (18 s rDNA sequences from 11 arbuscular mycorrhizal fungi (AMF) and 16 s rDNA sequences of 40 bacterial isolates from plant interior). The results indicate that the fuzzy integral algorithm is an efficient and feasible alignment-free method for sequence analysis on the genomic scale.

Protein Remote Homology Detection and Fold Recognition Based on Sequence-Order Frequency Matrix

Protein remote homology detection and fold recognition are two critical tasks for the studies of protein structures and functions. Currently, the profile-based methods achieve the state-of-the-art performance in these fields. However, the widely used sequence profiles, like position-specific frequency matrix (PSFM) and position-specific scoring matrix (PSSM), ignore the sequence-order effects along protein sequence. In this study, we have proposed a novel profile, called sequence-order frequency matrix (SOFM), to extract the sequence-order information of neighboring residues from multiple sequence alignment (MSA). Combined with two profile feature extraction approaches, top-n-grams and the Smith-Waterman algorithm, the SOFMs are applied to protein remote homology detection and fold recognition, and two predictors called SOFM-Top and SOFM-SW are proposed. Experimental results show that SOFM contains more information content than other profiles, and these two predictors outperform other state-of-the-art methods. It is anticipated that SOFM will become a very useful profile in the studies of protein structures and functions.

Protein remote homology detection based on bidirectional long short-term memory

BACKGROUND

Protein remote homology detection plays a vital role in studies of protein structures and functions. Almost all of the traditional machine leaning methods require fixed length features to represent the protein sequences. However, it is never an easy task to extract the discriminative features with limited knowledge of proteins. On the other hand, deep learning technique has demonstrated its advantage in automatically learning representations. It is worthwhile to explore the applications of deep learning techniques to the protein remote homology detection.

RESULTS

In this study, we employ the Bidirectional Long Short-Term Memory (BLSTM) to learn effective features from pseudo proteins, also propose a predictor called ProDec-BLSTM: it includes input layer, bidirectional LSTM, time distributed dense layer and output layer. This neural network can automatically extract the discriminative features by using bidirectional LSTM and the time distributed dense layer.

CONCLUSION

Experimental results on a widely-used benchmark dataset show that ProDec-BLSTM outperforms other related methods in terms of both the mean ROC and mean ROC50 scores. This promising result shows that ProDec-BLSTM is a useful tool for protein remote homology detection. Furthermore, the hidden patterns learnt by ProDec-BLSTM can be interpreted and visualized, and therefore, additional useful information can be obtained.

Fast and accurate phylogeny reconstruction using filtered spaced-word matches

Motivation

Word-based or ‘alignment-free’ algorithms are increasingly used for phylogeny reconstruction and genome comparison, since they are much faster than traditional approaches that are based on full sequence alignments. Existing alignment-free programs, however, are less accurate than alignment-based methods.

Results

We propose Filtered Spaced Word Matches (FSWM), a fast alignment-free approach to estimate phylogenetic distances between large genomic sequences. For a pre-defined binary pattern of match and don’t-care positions, FSWM rapidly identifies spaced word-matches between input sequences, i.e. gap-free local alignments with matching nucleotides at the match positions and with mismatches allowed at the don’t-care positions. We then estimate the number of nucleotide substitutions per site by considering the nucleotides aligned at the don’t-care positions of the identified spaced-word matches. To reduce the noise from spurious random matches, we use a filtering procedure where we discard all spaced-word matches for which the overall similarity between the aligned segments is below a threshold. We show that our approach can accurately estimate substitution frequencies even for distantly related sequences that cannot be analyzed with existing alignment-free methods; phylogenetic trees constructed with FSWM distances are of high quality. A program run on a pair of eukaryotic genomes of a few hundred Mb each takes a few minutes.

Availability and Implementation

The program source code for FSWM including a documentation, as well as the software that we used to generate artificial genome sequences are freely available at http://fswm.gobics.de/

Supplementary information

Supplementary data are available at Bioinformatics online.

Genetic sequence-based prediction of long-range chromatin interactions suggests a potential role of short tandem repeat sequences in genome organization

Background

Knowing the three-dimensional (3D) structure of the chromatin is important for obtaining a complete picture of the regulatory landscape. Changes in the 3D structure have been implicated in diseases. While there exist approaches that attempt to predict the long-range chromatin interactions, they focus only on interactions between specific genomic regions — the promoters and enhancers, neglecting other possibilities, for instance, the so-called structural interactions involving intervening chromatin.

Results

We present a method that can be trained on 5C data using the genetic sequence of the candidate loci to predict potential genome-wide interaction partners of a particular locus of interest. We have built locus-specific support vector machine (SVM)-based predictors using the oligomer distance histograms (ODH) representation. The method shows good performance with a mean test AUC (area under the receiver operating characteristic (ROC) curve) of 0.7 or higher for various regions across cell lines GM12878, K562 and HeLa-S3. In cases where any locus did not have sufficient candidate interaction partners for model training, we employed multitask learning to share knowledge between models of different loci. In this scenario, across the three cell lines, the method attained an average performance increase of 0.09 in the AUC. Performance evaluation of the models trained on 5C data regarding prediction on an independent high-resolution Hi-C dataset (which is a rather hard problem) shows 0.56 AUC, on average. Additionally, we have developed new, intuitive visualization methods that enable interpretation of sequence signals that contributed towards prediction of locus-specific interaction partners. The analysis of these sequence signals suggests a potential general role of short tandem repeat sequences in genome organization.

Conclusions

We demonstrated how our approach can 1) provide insights into sequence features of locus-specific interaction partners, and 2) also identify their cell-line specificity. That our models deem short tandem repeat sequences as discriminative for prediction of potential interaction partners, suggests that they could play a larger role in genome organization. Thus, our approach can (a) be beneficial to broadly understand, at the sequence-level, chromatin interactions and higher-order structures like (meta-) topologically associating domains (TADs); (b) study regions omitted from existing prediction approaches using various information sources (e.g., epigenetic information); and (c) improve methods that predict the 3D structure of the chromatin.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-017-1624-x) contains supplementary material, which is available to authorized users.

rasbhari: Optimizing Spaced Seeds for Database Searching, Read Mapping and Alignment-Free Sequence Comparison

Many algorithms for sequence analysis rely on word matching or word statistics. Often, these approaches can be improved if binary patterns representing match and don't-care positions are used as a filter, such that only those positions of words are considered that correspond to the match positions of the patterns. The performance of these approaches, however, depends on the underlying patterns. Herein, we show that the overlap complexity of a pattern set that was introduced by Ilie and Ilie is closely related to the variance of the number of matches between two evolutionarily related sequences with respect to this pattern set. We propose a modified hill-climbing algorithm to optimize pattern sets for database searching, read mapping and alignment-free sequence comparison of nucleic-acid sequences; our implementation of this algorithm is called rasbhari. Depending on the application at hand, rasbhari can either minimize the overlap complexity of pattern sets, maximize their sensitivity in database searching or minimize the variance of the number of pattern-based matches in alignment-free sequence comparison. We show that, for database searching, rasbhari generates pattern sets with slightly higher sensitivity than existing approaches. In our Spaced Words approach to alignment-free sequence comparison, pattern sets calculated with rasbhari led to more accurate estimates of phylogenetic distances than the randomly generated pattern sets that we previously used. Finally, we used rasbhari to generate patterns for short read classification with CLARK-S. Here too, the sensitivity of the results could be improved, compared to the default patterns of the program. We integrated rasbhari into Spaced Words; the source code of rasbhari is freely available at http://rasbhari.gobics.de/.

Protein Remote Homology Detection Based on an Ensemble Learning Approach

Protein remote homology detection is one of the central problems in bioinformatics. Although some computational methods have been proposed, the problem is still far from being solved. In this paper, an ensemble classifier for protein remote homology detection, called SVM-Ensemble, was proposed with a weighted voting strategy. SVM-Ensemble combined three basic classifiers based on different feature spaces, including Kmer, ACC, and SC-PseAAC. These features consider the characteristics of proteins from various perspectives, incorporating both the sequence composition and the sequence-order information along the protein sequences. Experimental results on a widely used benchmark dataset showed that the proposed SVM-Ensemble can obviously improve the predictive performance for the protein remote homology detection. Moreover, it achieved the best performance and outperformed other state-of-the-art methods.

Natural vs. random protein sequences: Discovering combinatorics properties on amino acid words

Casual mutations and natural selection have driven the evolution of protein amino acid sequences that we observe at present in nature. The question about which is the dominant force of proteins evolution is still lacking of an unambiguous answer. Casual mutations tend to randomize protein sequences while, in order to have the correct functionality, one expects that selection mechanisms impose rigid constraints on amino acid sequences. Moreover, one also has to consider that the space of all possible amino acid sequences is so astonishingly large that it could be reasonable to have a well tuned amino acid sequence indistinguishable from a random one. In order to study the possibility to discriminate between random and natural amino acid sequences, we introduce different measures of association between pairs of amino acids in a sequence, and apply them to a dataset of 1047 natural protein sequences and 10,470 random sequences, carefully generated in order to preserve the relative length and amino acid distribution of the natural proteins. We analyze the multidimensional measures with machine learning techniques and show that, to a reasonable extent, natural protein sequences can be differentiated from random ones.

Estimating evolutionary distances between genomic sequences from spaced-word matches

Alignment-free methods are increasingly used to calculate evolutionary distances between DNA and protein sequences as a basis of phylogeny reconstruction. Most of these methods, however, use heuristic distance functions that are not based on any explicit model of molecular evolution. Herein, we propose a simple estimator d N of the evolutionary distance between two DNA sequences that is calculated from the number N of (spaced) word matches between them. We show that this distance function is more accurate than other distance measures that are used by alignment-free methods. In addition, we calculate the variance of the normalized number N of (spaced) word matches. We show that the variance of N is smaller for spaced words than for contiguous words, and that the variance is further reduced if our spaced-words approach is used with multiple patterns of 'match positions' and 'don't care positions'. Our software is available online and as downloadable source code at: http://spaced.gobics.de/.

Kmacs: the k-mismatch average common substring approach to alignment-free sequence comparison

MOTIVATION

Alignment-based methods for sequence analysis have various limitations if large datasets are to be analysed. Therefore, alignment-free approaches have become popular in recent years. One of the best known alignment-free methods is the average common substring approach that defines a distance measure on sequences based on the average length of longest common words between them. Herein, we generalize this approach by considering longest common substrings with k mismatches. We present a greedy heuristic to approximate the length of such k-mismatch substrings, and we describe kmacs, an efficient implementation of this idea based on generalized enhanced suffix arrays.

RESULTS

To evaluate the performance of our approach, we applied it to phylogeny reconstruction using a large number of DNA and protein sequence sets. In most cases, phylogenetic trees calculated with kmacs were more accurate than trees produced with established alignment-free methods that are based on exact word matches. Especially on protein sequences, our method seems to be superior. On simulated protein families, kmacs even outperformed a classical approach to phylogeny reconstruction using multiple alignment and maximum likelihood.

AVAILABILITY AND IMPLEMENTATION

kmacs is implemented in C++, and the source code is freely available at http://kmacs.gobics.de/.

Fast alignment-free sequence comparison using spaced-word frequencies

Motivation: Alignment-free methods for sequence comparison are increasingly used for genome analysis and phylogeny reconstruction; they circumvent various difficulties of traditional alignment-based approaches. In particular, alignment-free methods are much faster than pairwise or multiple alignments. They are, however, less accurate than methods based on sequence alignment. Most alignment-free approaches work by comparing the word composition of sequences. A well-known problem with these methods is that neighbouring word matches are far from independent.

Results: To reduce the statistical dependency between adjacent word matches, we propose to use ‘spaced words’, defined by patterns of ‘match’ and ‘don’t care’ positions, for alignment-free sequence comparison. We describe a fast implementation of this approach using recursive hashing and bit operations, and we show that further improvements can be achieved by using multiple patterns instead of single patterns. To evaluate our approach, we use spaced-word frequencies as a basis for fast phylogeny reconstruction. Using real-world and simulated sequence data, we demonstrate that our multiple-pattern approach produces better phylogenies than approaches relying on contiguous words.

Availability and implementation: Our program is freely available at http://spaced.gobics.de/.

Contact:chris.leimeister@stud.uni-goettingen.de

Supplementary information:Supplementary data are available at Bioinformatics online.

Using distances between Top-n-gram and residue pairs for protein remote homology detection

Background

Protein remote homology detection is one of the central problems in bioinformatics, which is important for both basic research and practical application. Currently, discriminative methods based on Support Vector Machines (SVMs) achieve the state-of-the-art performance. Exploring feature vectors incorporating the position information of amino acids or other protein building blocks is a key step to improve the performance of the SVM-based methods.

Results

Two new methods for protein remote homology detection were proposed, called SVM-DR and SVM-DT. SVM-DR is a sequence-based method, in which the feature vector representation for protein is based on the distances between residue pairs. SVM-DT is a profile-based method, which considers the distances between Top-n-gram pairs. Top-n-gram can be viewed as a profile-based building block of proteins, which is calculated from the frequency profiles. These two methods are position dependent approaches incorporating the sequence-order information of protein sequences. Various experiments were conducted on a benchmark dataset containing 54 families and 23 superfamilies. Experimental results showed that these two new methods are very promising. Compared with the position independent methods, the performance improvement is obvious. Furthermore, the proposed methods can also provide useful insights for studying the features of protein families.

Conclusion

The better performance of the proposed methods demonstrates that the position dependant approaches are efficient for protein remote homology detection. Another advantage of our methods arises from the explicit feature space representation, which can be used to analyze the characteristic features of protein families. The source code of SVM-DT and SVM-DR is available at http://bioinformatics.hitsz.edu.cn/DistanceSVM/index.jsp

Combining evolutionary information extracted from frequency profiles with sequence-based kernels for protein remote homology detection

Motivation: Owing to its importance in both basic research (such as molecular evolution and protein attribute prediction) and practical application (such as timely modeling the 3D structures of proteins targeted for drug development), protein remote homology detection has attracted a great deal of interest. It is intriguing to note that the profile-based approach is promising and holds high potential in this regard. To further improve protein remote homology detection, a key step is how to find an optimal means to extract the evolutionary information into the profiles.

Results: Here, we propose a novel approach, the so-called profile-based protein representation, to extract the evolutionary information via the frequency profiles. The latter can be calculated from the multiple sequence alignments generated by PSI-BLAST. Three top performing sequence-based kernels (SVM-Ngram, SVM-pairwise and SVM-LA) were combined with the profile-based protein representation. Various tests were conducted on a SCOP benchmark dataset that contains 54 families and 23 superfamilies. The results showed that the new approach is promising, and can obviously improve the performance of the three kernels. Furthermore, our approach can also provide useful insights for studying the features of proteins in various families. It has not escaped our notice that the current approach can be easily combined with the existing sequence-based methods so as to improve their performance as well.

Availability and implementation: For users’ convenience, the source code of generating the profile-based proteins and the multiple kernel learning was also provided at

http://bioinformatics.hitsz.edu.cn/main/∼binliu/remote/

Contact:bliu@insun.hit.edu.cn or bliu@gordonlifescience.org

Supplementary information:Supplementary data are available at Bioinformatics online.

Using amino acid physicochemical distance transformation for fast protein remote homology detection

Protein remote homology detection is one of the most important problems in bioinformatics. Discriminative methods such as support vector machines (SVM) have shown superior performance. However, the performance of SVM-based methods depends on the vector representations of the protein sequences. Prior works have demonstrated that sequence-order effects are relevant for discrimination, but little work has explored how to incorporate the sequence-order information along with the amino acid physicochemical properties into the prediction. In order to incorporate the sequence-order effects into the protein remote homology detection, the physicochemical distance transformation (PDT) method is proposed. Each protein sequence is converted into a series of numbers by using the physicochemical property scores in the amino acid index (AAIndex), and then the sequence is converted into a fixed length vector by PDT. The sequence-order information can be efficiently included into the feature vector with little computational cost by this approach. Finally, the feature vectors are input into a support vector machine classifier to detect the protein remote homologies. Our experiments on a well-known benchmark show the proposed method SVM-PDT achieves superior or comparable performance with current state-of-the-art methods and its computational cost is considerably superior to those of other methods. When the evolutionary information extracted from the frequency profiles is combined with the PDT method, the profile-based PDT approach can improve the performance by 3.4% and 11.4% in terms of ROC score and ROC50 score respectively. The local sequence-order information of the protein can be efficiently captured by the proposed PDT and the physicochemical properties extracted from the amino acid index are incorporated into the prediction. The physicochemical distance transformation provides a general framework, which would be a valuable tool for protein-level study.

Remote protein homology detection and fold recognition using two-layer support vector machine classifiers

Remote protein homology detection and fold recognition refer to detection of structural homology in proteins where there are small or no similarities in the sequence. To detect protein structural classes from protein primary sequence information, homology-based methods have been developed, which can be divided to three types: discriminative classifiers, generative models for protein families and pairwise sequence comparisons. Support Vector Machines (SVM) and Neural Networks (NN) are two popular discriminative methods. Recent studies have shown that SVM has fast speed during training, more accurate and efficient compared to NN. We present a comprehensive method based on two-layer classifiers. The 1st layer is used to detect up to superfamily and family in SCOP hierarchy using optimized binary SVM classification rules. It used the kernel function known as the Bio-kernel, which incorporates the biological information in the classification process. The 2nd layer uses discriminative SVM algorithm with string kernel that will detect up to protein fold level in SCOP hierarchy. The results obtained were evaluated using mean ROC and mean MRFP and the significance of the result produced with pairwise t-test was tested. Experimental results show that our approaches significantly improve the performance of remote protein homology detection and fold recognition for all three different version SCOP datasets (1.53, 1.67 and 1.73). We achieved 4.19% improvements in term of mean ROC in SCOP 1.53, 4.75% in SCOP 1.67 and 4.03% in SCOP 1.73 datasets when compared to the result produced by well-known methods. The combination of first layer and second layer of BioSVM-2L performs well in remote homology detection and fold recognition even in three different versions of datasets.

A discriminative method for family-based protein remote homology detection that combines inductive logic programming and propositional models

Background

Remote homology detection is a hard computational problem. Most approaches have trained computational models by using either full protein sequences or multiple sequence alignments (MSA), including all positions. However, when we deal with proteins in the "twilight zone" we can observe that only some segments of sequences (motifs) are conserved. We introduce a novel logical representation that allows us to represent physico-chemical properties of sequences, conserved amino acid positions and conserved physico-chemical positions in the MSA. From this, Inductive Logic Programming (ILP) finds the most frequent patterns (motifs) and uses them to train propositional models, such as decision trees and support vector machines (SVM).

Results

We use the SCOP database to perform our experiments by evaluating protein recognition within the same superfamily. Our results show that our methodology when using SVM performs significantly better than some of the state of the art methods, and comparable to other. However, our method provides a comprehensible set of logical rules that can help to understand what determines a protein function.

Conclusions

The strategy of selecting only the most frequent patterns is effective for the remote homology detection. This is possible through a suitable first-order logical representation of homologous properties, and through a set of frequent patterns, found by an ILP system, that summarizes essential features of protein functions.

MS4--Multi-Scale Selector of Sequence Signatures: an alignment-free method for classification of biological sequences

BACKGROUND

While multiple alignment is the first step of usual classification schemes for biological sequences, alignment-free methods are being increasingly used as alternatives when multiple alignments fail. Subword-based combinatorial methods are popular for their low algorithmic complexity (suffix trees ...) or exhaustivity (motif search), in general with fixed length word and/or number of mismatches. We developed previously a method to detect local similarities (the N-local decoding) based on the occurrences of repeated subwords of fixed length, which does not impose a fixed number of mismatches. The resulting similarities are, for some "good" values of N, sufficiently relevant to form the basis of a reliable alignment-free classification. The aim of this paper is to develop a method that uses the similarities detected by N-local decoding while not imposing a fixed value of N. We present a procedure that selects for every position in the sequences an adaptive value of N, and we implement it as the MS4 classification tool.

RESULTS

Among the equivalence classes produced by the N-local decodings for all N, we select a (relatively) small number of "relevant" classes corresponding to variable length subwords that carry enough information to perform the classification. The parameter N, for which correct values are data-dependent and thus hard to guess, is here replaced by the average repetitivity kappa of the sequences. We show that our approach yields classifications of several sets of HIV/SIV sequences that agree with the accepted taxonomy, even on usually discarded repetitive regions (like the non-coding part of LTR).

CONCLUSIONS

The method MS4 satisfactorily classifies a set of sequences that are notoriously hard to align. This suggests that our approach forms the basis of a reliable alignment-free classification tool. The only parameter kappa of MS4 seems to give reasonable results even for its default value, which can be a great advantage for sequence sets for which little information is available.

Physicochemical property distributions for accurate and rapid pairwise protein homology detection

Background

The challenge of remote homology detection is that many evolutionarily related sequences have very little similarity at the amino acid level. Kernel-based discriminative methods, such as support vector machines (SVMs), that use vector representations of sequences derived from sequence properties have been shown to have superior accuracy when compared to traditional approaches for the task of remote homology detection.

Results

We introduce a new method for feature vector representation based on the physicochemical properties of the primary protein sequence. A distribution of physicochemical property scores are assembled from 4-mers of the sequence and normalized based on the null distribution of the property over all possible 4-mers. With this approach there is little computational cost associated with the transformation of the protein into feature space, and overall performance in terms of remote homology detection is comparable with current state-of-the-art methods. We demonstrate that the features can be used for the task of pairwise remote homology detection with improved accuracy versus sequence-based methods such as BLAST and other feature-based methods of similar computational cost.

Conclusions

A protein feature method based on physicochemical properties is a viable approach for extracting features in a computationally inexpensive manner while retaining the sensitivity of SVM protein homology detection. Furthermore, identifying features that can be used for generic pairwise homology detection in lieu of family-based homology detection is important for applications such as large database searches and comparative genomics.

UFO: a web server for ultra-fast functional profiling of whole genome protein sequences

Background

Functional profiling is a key technique to characterize and compare the functional potential of entire genomes. The estimation of profiles according to an assignment of sequences to functional categories is a computationally expensive task because it requires the comparison of all protein sequences from a genome with a usually large database of annotated sequences or sequence families.

Description

Based on machine learning techniques for Pfam domain detection, the UFO web server for ultra-fast functional profiling allows researchers to process large protein sequence collections instantaneously. Besides the frequencies of Pfam and GO categories, the user also obtains the sequence specific assignments to Pfam domain families. In addition, a comparison with existing genomes provides dissimilarity scores with respect to 821 reference proteomes. Considering the underlying UFO domain detection, the results on 206 test genomes indicate a high sensitivity of the approach. In comparison with current state-of-the-art HMMs, the runtime measurements show a considerable speed up in the range of four orders of magnitude. For an average size prokaryotic genome, the computation of a functional profile together with its comparison typically requires about 10 seconds of processing time.

Conclusion

For the first time the UFO web server makes it possible to get a quick overview on the functional inventory of newly sequenced organisms. The genome scale comparison with a large number of precomputed profiles allows a first guess about functionally related organisms. The service is freely available and does not require user registration or specification of a valid email address.

HIV-1 coreceptor usage prediction without multiple alignments: an application of string kernels

BACKGROUND

Human immunodeficiency virus type 1 (HIV-1) infects cells by means of ligand-receptor interactions. This lentivirus uses the CD4 receptor in conjunction with a chemokine coreceptor, either CXCR4 or CCR5, to enter a target cell. HIV-1 is characterized by high sequence variability. Nonetheless, within this extensive variability, certain features must be conserved to define functions and phenotypes. The determination of coreceptor usage of HIV-1, from its protein envelope sequence, falls into a well-studied machine learning problem known as classification. The support vector machine (SVM), with string kernels, has proven to be very efficient for dealing with a wide class of classification problems ranging from text categorization to protein homology detection. In this paper, we investigate how the SVM can predict HIV-1 coreceptor usage when it is equipped with an appropriate string kernel.

RESULTS

Three string kernels were compared. Accuracies of 96.35% (CCR5) 94.80% (CXCR4) and 95.15% (CCR5 and CXCR4) were achieved with the SVM equipped with the distant segments kernel on a test set of 1425 examples with a classifier built on a training set of 1425 examples. Our datasets are built with Los Alamos National Laboratory HIV Databases sequences. A web server is available at http://genome.ulaval.ca/hiv-dskernel.

CONCLUSION

We examined string kernels that have been used successfully for protein homology detection and propose a new one that we call the distant segments kernel. We also show how to extract the most relevant features for HIV-1 coreceptor usage. The SVM with the distant segments kernel is currently the best method described.

A discriminative method for protein remote homology detection and fold recognition combining Top-n-grams and latent semantic analysis

Background

Protein remote homology detection and fold recognition are central problems in bioinformatics. Currently, discriminative methods based on support vector machine (SVM) are the most effective and accurate methods for solving these problems. A key step to improve the performance of the SVM-based methods is to find a suitable representation of protein sequences.

Results

In this paper, a novel building block of proteins called Top-n-grams is presented, which contains the evolutionary information extracted from the protein sequence frequency profiles. The protein sequence frequency profiles are calculated from the multiple sequence alignments outputted by PSI-BLAST and converted into Top-n-grams. The protein sequences are transformed into fixed-dimension feature vectors by the occurrence times of each Top-n-gram. The training vectors are evaluated by SVM to train classifiers which are then used to classify the test protein sequences. We demonstrate that the prediction performance of remote homology detection and fold recognition can be improved by combining Top-n-grams and latent semantic analysis (LSA), which is an efficient feature extraction technique from natural language processing. When tested on superfamily and fold benchmarks, the method combining Top-n-grams and LSA gives significantly better results compared to related methods.

Conclusion

The method based on Top-n-grams significantly outperforms the methods based on many other building blocks including N-grams, patterns, motifs and binary profiles. Therefore, Top-n-gram is a good building block of the protein sequences and can be widely used in many tasks of the computational biology, such as the sequence alignment, the prediction of domain boundary, the designation of knowledge-based potentials and the prediction of protein binding sites.

MotifCluster: an interactive online tool for clustering and visualizing sequences using shared motifs

MotifCluster finds related motifs in a set of sequences and clusters the sequences into families using the motifs they contain.

MotifCluster finds related motifs in a set of sequences, and clusters the sequences into families using the motifs they contain. MotifCluster, at , lets users test whether proteins are related, cluster sequences by shared conserved motifs, and visualize motifs mapped onto trees, sequences and three-dimensional structures. We demonstrate MotifCluster's accuracy using gold-standard protein superfamilies; using recommended settings, families were assigned to the correct superfamilies with 0.17% false positive and no false negative assignments.

Word correlation matrices for protein sequence analysis and remote homology detection

Background

Classification of protein sequences is a central problem in computational biology. Currently, among computational methods discriminative kernel-based approaches provide the most accurate results. However, kernel-based methods often lack an interpretable model for analysis of discriminative sequence features, and predictions on new sequences usually are computationally expensive.

Results

In this work we present a novel kernel for protein sequences based on average word similarity between two sequences. We show that this kernel gives rise to a feature space that allows analysis of discriminative features and fast classification of new sequences. We demonstrate the performance of our approach on a widely-used benchmark setup for protein remote homology detection.

Conclusion

Our word correlation approach provides highly competitive performance as compared with state-of-the-art methods for protein remote homology detection. The learned model is interpretable in terms of biologically meaningful features. In particular, analysis of discriminative words allows the identification of characteristic regions in biological sequences. Because of its high computational efficiency, our method can be applied to ranking of potential homologs in large databases.

Probabilistic multi-class multi-kernel learning: on protein fold recognition and remote homology detection

MOTIVATION

The problems of protein fold recognition and remote homology detection have recently attracted a great deal of interest as they represent challenging multi-feature multi-class problems for which modern pattern recognition methods achieve only modest levels of performance. As with many pattern recognition problems, there are multiple feature spaces or groups of attributes available, such as global characteristics like the amino-acid composition (C), predicted secondary structure (S), hydrophobicity (H), van der Waals volume (V), polarity (P), polarizability (Z), as well as attributes derived from local sequence alignment such as the Smith-Waterman scores. This raises the need for a classification method that is able to assess the contribution of these potentially heterogeneous object descriptors while utilizing such information to improve predictive performance. To that end, we offer a single multi-class kernel machine that informatively combines the available feature groups and, as is demonstrated in this article, is able to provide the state-of-the-art in performance accuracy on the fold recognition problem. Furthermore, the proposed approach provides some insight by assessing the significance of recently introduced protein features and string kernels. The proposed method is well-founded within a Bayesian hierarchical framework and a variational Bayes approximation is derived which allows for efficient CPU processing times.

RESULTS

The best performance which we report on the SCOP PDB-40D benchmark data-set is a 70% accuracy by combining all the available feature groups from global protein characteristics but also including sequence-alignment features. We offer an 8% improvement on the best reported performance that combines multi-class k-nn classifiers while at the same time reducing computational costs and assessing the predictive power of the various available features. Furthermore, we examine the performance of our methodology on the SCOP 1.53 benchmark data-set that simulates remote homology detection and examine the combination of various state-of-the-art string kernels that have recently been proposed.

Statistical learning of peptide retention behavior in chromatographic separations: a new kernel-based approach for computational proteomics

Background

High-throughput peptide and protein identification technologies have benefited tremendously from strategies based on tandem mass spectrometry (MS/MS) in combination with database searching algorithms. A major problem with existing methods lies within the significant number of false positive and false negative annotations. So far, standard algorithms for protein identification do not use the information gained from separation processes usually involved in peptide analysis, such as retention time information, which are readily available from chromatographic separation of the sample. Identification can thus be improved by comparing measured retention times to predicted retention times. Current prediction models are derived from a set of measured test analytes but they usually require large amounts of training data.

Results

We introduce a new kernel function which can be applied in combination with support vector machines to a wide range of computational proteomics problems. We show the performance of this new approach by applying it to the prediction of peptide adsorption/elution behavior in strong anion-exchange solid-phase extraction (SAX-SPE) and ion-pair reversed-phase high-performance liquid chromatography (IP-RP-HPLC). Furthermore, the predicted retention times are used to improve spectrum identifications by a p-value-based filtering approach. The approach was tested on a number of different datasets and shows excellent performance while requiring only very small training sets (about 40 peptides instead of thousands). Using the retention time predictor in our retention time filter improves the fraction of correctly identified peptide mass spectra significantly.

Conclusion

The proposed kernel function is well-suited for the prediction of chromatographic separation in computational proteomics and requires only a limited amount of training data. The performance of this new method is demonstrated by applying it to peptide retention time prediction in IP-RP-HPLC and prediction of peptide sample fractionation in SAX-SPE. Finally, we incorporate the predicted chromatographic behavior in a p-value based filter to improve peptide identifications based on liquid chromatography-tandem mass spectrometry.

Fast model-based protein homology detection without alignment

MOTIVATION

As more genomes are sequenced, the demand for fast gene classification techniques is increasing. To analyze a newly sequenced genome, first the genes are identified and translated into amino acid sequences which are then classified into structural or functional classes. The best-performing protein classification methods are based on protein homology detection using sequence alignment methods. Alignment methods have recently been enhanced by discriminative methods like support vector machines (SVMs) as well as by position-specific scoring matrices (PSSM) as obtained from PSI-BLAST. However, alignment methods are time consuming if a new sequence must be compared to many known sequences-the same holds for SVMs. Even more time consuming is to construct a PSSM for the new sequence. The best-performing methods would take about 25 days on present-day computers to classify the sequences of a new genome (20,000 genes) as belonging to just one specific class--however, there are hundreds of classes. Another shortcoming of alignment algorithms is that they do not build a model of the positive class but measure the mutual distance between sequences or profiles. Only multiple alignments and hidden Markov models are popular classification methods which build a model of the positive class but they show low classification performance. The advantage of a model is that it can be analyzed for chemical properties common to the class members to obtain new insights into protein function and structure. We propose a fast model-based recurrent neural network for protein homology detection, the 'Long Short-Term Memory' (LSTM). LSTM automatically extracts indicative patterns for the positive class, but in contrast to profile methods it also extracts negative patterns and uses correlations between all detected patterns for classification. LSTM is capable to automatically extract useful local and global sequence statistics like hydrophobicity, polarity, volume, polarizability and combine them with a pattern. These properties make LSTM complementary to alignment-based approaches as it does not use predefined similarity measures like BLOSUM or PAM matrices.

RESULTS

We have applied LSTM to a well known benchmark for remote protein homology detection, where a protein must be classified as belonging to a SCOP superfamily. LSTM reaches state-of-the-art classification performance but is considerably faster for classification than other approaches with comparable classification performance. LSTM is five orders of magnitude faster than methods which perform slightly better in classification and two orders of magnitude faster than the fastest SVM-based approaches (which, however, have lower classification performance than LSTM). Only PSI-BLAST and HMM-based methods show comparable time complexity as LSTM, but they cannot compete with LSTM in classification performance. To test the modeling capabilities of LSTM, we applied LSTM to PROSITE classes and interpreted the extracted patterns. In 8 out of 15 classes, LSTM automatically extracted the PROSITE motif. In the remaining 7 cases alternative motifs are generated which give better classification results on average than the PROSITE motifs.

AVAILABILITY

The LSTM algorithm is available from http://www.bioinf.jku.at/software/LSTM_protein/.