 Proceedings
 Open Access
Protein sequence classification using feature hashing
 Cornelia Caragea^{1}Email author,
 Adrian Silvescu^{2} and
 Prasenjit Mitra^{1, 2}
https://doi.org/10.1186/1477595610S1S14
© Caragea et al; licensee BioMed Central Ltd. 2012
 Published: 21 June 2012
Abstract
Recent advances in nextgeneration sequencing technologies have resulted in an exponential increase in the rate at which protein sequence data are being acquired. The kgram feature representation, commonly used for protein sequence classification, usually results in prohibitively high dimensional input spaces, for large values of k. Applying data mining algorithms to these input spaces may be intractable due to the large number of dimensions. Hence, using dimensionality reduction techniques can be crucial for the performance and the complexity of the learning algorithms. In this paper, we study the applicability of feature hashing to protein sequence classification, where the original highdimensional space is "reduced" by hashing the features into a lowdimensional space, using a hash function, i.e., by mapping features into hash keys, where multiple features can be mapped (at random) to the same hash key, and "aggregating" their counts. We compare feature hashing with the "bag of kgrams" approach. Our results show that feature hashing is an effective approach to reducing dimensionality on protein sequence classification tasks.
Keywords
 Hash Function
 Support Vector Machine Classifier
 Latent Dirichlet Allocation
 Average Mutual Information
 Latent Semantic Indexing
Introduction
Many problems in computational biology, e.g., protein function prediction, subcellular localization prediction, etc., can be formulated as sequence classification tasks [1], where the amino acid sequence of a protein is used to classify the protein in functional and localization classes.
Protein sequence data contain intrinsic dependencies between their constituent elements. Given a protein sequence x = x _{0}, ⋯, x _{ n1}over the amino acid alphabet, the dependencies between neighboring elements can be modeled by generating all the contiguous (potentially overlapping) subsequences of a certain length k, x _{ ik }, ⋯, x _{ i1}, i = k, ⋯, n, called kgrams, or sequence motifs. Because the protein sequence motifs may have variable lengths, generating the kgrams can be done by sliding a window of length k over the sequence x, for various values of k. Exploiting dependencies in the data increases the richness of the representation. However, the fixed or variable length kgram representations, used for protein sequence classification, usually result in prohibitively highdimensional input spaces, for large values of k. Applying data mining algorithms to these input spaces may be intractable due to the large number of dimensions. Hence, using dimensionality reduction techniques can be crucial for the performance and the complexity of the learning algorithms.
In this paper, we study the applicability of feature hashing to protein sequence classification and address the following main questions: (i) How effective is feature hashing on prohibitively high dimensional kgram representations?; (ii) What is the influence of the hash size (i.e., the reduced dimension) on the performance of protein sequence classifiers that use hash features, and what is the hash size at which the performance starts degrading, due to hash collisions?; and (iii) How does the performance of feature hashing compare to that of the "bag of kgrams" approach? The results of our experiments on three protein subcellular localization data sets show that feature hashing is effective at reducing dimensionality on protein sequence classification tasks.
The paper is organized as follows. In Section 2, we discuss the related work. We provide background on feature hashing in Section 3. Section 4 presents experiments and results, and Section 5 concludes the paper.
Related work
Feature selection
Feature selection [5, 7, 12] is a dimensionality reduction technique, which attempts to remove redundant or irrelevant features in order to improve classification performance of learning algorithms. Feature selection methods have been widely used in Bioinformatics for tasks such as protein function prediction and gene prediction, where the features could be kgrams; microarray analysis; mass spectra analysis; single nucleotide polymorphisms (SNPs) analysis, among others (see [13] for a review).
Topic models
Topic models, such as Latent Dirichlet Allocation (LDA) [3], Probabilistic Latent Semantic Analysis (PLSA) [4], and Latent Semantic Indexing (LSI) [14] are dimensionality reduction models, designed to uncover hidden topics, i.e., clusters of semantically related words that cooccur in text documents. LSI uses singular value decomposition to identify topics, which are then used to represent documents in a low dimensional "topic" space. LDA models each document as a mixture of topics (drawn from a conjugate Dirichlet prior), and each topic as a distribution over the words in the vocabulary. LDA has recently emerged as an important tool for modeling protein data. For example, Airoldi et al. [15] proposed the mixed membership stochastic block models to learn hidden protein interaction patterns. Pan et al. [16] used LDA to discover latent topic features, which correspond to hidden structures in the protein data, and input these features to random forest classifiers to predict protein interactions. However, topic models are computationally expensive, for example, LDA requires inference at runtime to estimate the topic distribution.
Feature abstraction
Feature abstraction methods [17] are designed to reduce a model input size by grouping "similar" features into clusters of features. Specifically, feature abstraction learns an abstraction hierarchy over the set of features using hierarchical agglomerative clustering, based on the JensenShannon divergence. A cut or level of abstraction through the resulting abstraction hierarchy specifies a compressed model, where the nodes (or abstractions) on the cut are used as "features" in a classification model. Silvescu et al. [17] used feature abstraction to simplify the data representation provided to a learner on biological sequence classification tasks.
Feature hashing
Shi et al. [8] and Weinberger et al. [9] presented hash kernels to map the highdimensional input spaces into lowdimensional spaces for large scale classification and large scale multitask learning (i.e., personalized spam filtering for hundreds of thousands of users), respectively. Ganchev and Dredze [18] empirically showed that hash features can produce accurate results on various NLP applications. Forman and Kirshenbaum [10] proposed a fast feature extraction approach by combining parsing and hashing for text classification and indexing. Hashing techniques have been also used in Bioinformatics. For example, Wesselink et al. [19] applied hashing to find the shortest contiguous subsequence that uniquely identifies a DNA sequence from a collection of DNA sequences. Buhler and Tompa [20] applied LocalitySensitive Hashing (LSH) [21], a random hashing/projection technique, to discover transcriptional regulatory motifs in eukaryotes and ribosome binding sites in prokaryotes. Furthermore, Buhler [22] applied LSH to find short ungapped local alignments on a genomewide scale. Shi et al. [8] used hashing to compare all subgraph pairs on biological graphs.
Markov models
In the context of protein sequence classification, it is worth mentioning the fixed and variableorder Markov models (MMs), which capture dependencies between neighboring sequence elements. MMs are among the most widely used generative models of sequence data [23]. In a k^{th} order MM, the sequence elements satisfy the Markov property: each element is independent of the rest given the k preceding elements. One main disadvantage of MMs in practice is that the number of parameters increases exponentially with the range k of direct dependencies, thereby increasing the risk of overfitting. Begleiter et al. [24] (and papers cited therein) have examined methods for prediction using variable order MMs, including probabilistic suffix trees, which can be viewed as variants of abstraction wherein the abstractions are constrained to share suffixes.
In contrast to the approaches above, we used feature hashing, a very inexpensive approach, to reduce dimensionality on protein sequence classification tasks, and compared it with the "bag of kgrams" approach.
Methods
The traditional kgram approaches construct a vocabulary of size d, which contains all kgrams in a protein data set. A protein sequence is represented as a vector x with as many entries as the number of kgrams in the vocabulary. For a protein sequence, an entry i in x can record the frequency of kgram i in the sequence, denoted by x _{ i }. Because only a small number of kgrams (compared to the vocabulary size) occur in a particular sequence, the representation of x is very sparse, i.e., only a small number of entries of x are nonzero. However, storing the parameter vectors in the original input space requires O(d) numbers, which
Algorithm 1 Feature Hashing
Input: Protein sequence x; hash functions h and ξ, $h:\mathcal{S}\to \left\{0,\cdots \phantom{\rule{0.3em}{0ex}},b1\right\},\xi :\mathcal{S}\to \left\{\pm 1\right\}$.
Output: Hashed feature vector x^{ h }.
x^{ h }: = [0, ⋯, 0];
for all kgram ∈ x do
i = h (kgram) % b; //Places kgrams into hash bins, from 0 to b1.
${x}_{i}^{h}={x}_{i}^{h}+\xi \left(k\mathsf{\text{gram}}\right)$; //Updates the i^{ th }hash feature value.
end for
return x^{ h }//Records values of hash features.
may become difficult given today's very large collections of protein and DNA sequence data. Feature hashing eliminates the need for such a requirement by implicitly encoding the mapping into a hash function. Next, we briefly overview feature hashing.
Feature hashing
Feature hashing [8–11] is a dimensionality reduction technique, in which highdimensional input vectors x of size d are hashed into lowdimensional feature vectors x^{ h }of size b. The procedure for hashing a protein sequence x is shown in Algorithm 1 and is briefly described next (see also Figure 1). Let $\mathcal{S}$ denote the set of all possible strings (or kgrams) and h and ξ be two hash functions, such that $h:\mathcal{S}\to \left\{0,\cdots \phantom{\rule{0.3em}{0ex}},b1\right\}$ and $\xi :\mathcal{S}\to \left\{\pm 1\right\}$, respectively. For a protein sequence x, each kgram in x is directly mapped, using h, into a hash key, which represents the index of the kgram in the feature vector x^{ h }, such that the hash key is a number between 0 and b  1. Note that h can be any hash function, e.g. hashCode() of the Java String class, or murmurHash function available online at http://sites.google.com/site/murmurhash/. Each index in x^{ h }stores the value ("frequency counts") of the corresponding hash feature. The hash function ξ indicates whether to increment or decrement the hash dimension of the kgram, which renders the hash feature vector x^{ h }to be unbiased (see [9] for more details).
However, for many practical applications, the value of b can be smaller than the theoretical lower bound. This may be problematic as the smaller the size of the hash vector x^{ h }becomes, the more collisions occur in the data. Even a single collision of very high frequency words with different class distributions, can result in significant loss of information. Next, we empirically study the applicability of feature hashing on a protein subcellular localization prediction task.
Experiments and results
We used three protein subcellular localization data sets in our study: psortNeg introduced in [25] and available online at http://www.psort.org/dataset/datasetv2.html, and plant, and nonplant introduced in [26] and available online at http://www.cbs.dtu.dk/services/TargetP/datasets/datasets.php. The psortNeg data set is extracted from PSORTdb v.2.0 Gramnegative sequences, which contains experimentally verified localization sites. Our data set consists of all proteins that belong to exactly one of the following five classes: cytoplasm (278), cytoplasmic membrane (309), periplasm (276), outer membrane (391) and extracellular (190). The total number of proteins in this data set is 1444. The plant data set contains 940 proteins belonging to one of the following four classes: chloroplast (141), mitochondrial (368), secretory pathway/signal peptide (269) and other (consisting of 54 proteins with label nuclear and 108 examples with label cytosolic). The nonplant data set contains 2738 proteins, each in one of the following three classes: mitochondrial (361), secretory pathway/signal peptide (715) and other (consisting of 1224 proteins labeled nuclear and 438 proteins labeled cytosolic).
Experimental design
Our experiments are designed to explore the following questions: (i) How effective is feature hashing on prohibitively highdimensional kgram representations?; (ii) What is the influence of the hash size on the performance of biological sequence classifiers that use hash features, and what is the hash size at which the performance starts degrading, due to hash collisions?; and (iii) How does the performance of feature hashing compare to that of the "bag of kgrams" approach?
To answer these questions, we proceeded with the following steps. We first preprocessed the data by generating all the kgrams from each collection of sequences, i.e., generating all the contiguous (potentially overlapping) subsequences of length k, for various values of k. This was done by sliding a window of length k over sequences in each data set. Note that if a kgram does not appear in the data, it was not considered as a feature. The number of unique kgrams is exponential in k. However, for large values of k, many of the kgrams may not appear in the data (and, consequently, their frequency counts are zero).
Given a protein sequence x, we applied feature hashing in two settings as follows: (i) We first generated all the kgrams of a fixed length k, where k = 3. Each such kgram was then hashed into a hash key. We refer to this setting as the fixedlength kgrams; (ii) We then generated all the kgrams of various lengths k, for values of k = 1, 2, 3, and 4. Thus, this setting uses the union of kgrams, for values of k ranging from 1 to 4. Each such kgram (i.e., unigram, 2gram, 3gram, or 4gram) was hashed into a hash key. We refer to this setting as the variablelength kgrams.
We trained Support Vector Machine (SVM) classifiers [27] on hash features, in both settings, fixedlength and variablelength kgrams, and investigated the influence of the hash size on the performance of the classifiers. Specifically, we trained SVM classifiers for values of the hash size (i.e., the reduced dimension) ranging from 2^{10} to 2^{22}, in steps of 1 for the powers of 2, and compared their performance.
Furthermore, we applied feature hashing to sparse highdimensional variablelength kgram representations to reduce the dimensionality to a midsize bdimensional space, e.g., b = 2^{16} or b = 2^{14}, and compared the performance of SVM classifiers trained using hash features with that of SVM classifiers trained using "bag of kgrams".
Specifically, the feature representations used in each case are the following:

a bag of d variablelength kgrams (where all the variablelength kgrams are used as features). This experiment is denoted by baseline.

a bag of b hash features obtained using feature hashing over all d variablelength kgrams, i.e., for each kgram, feature hashing produces an index i such that i = h(kgram) % b, where h represents a hash function. This experiment is denoted by FH.
In our experiments, we used the LibLinear implementation of SVM, available at http://www.csie.ntu.edu.tw/~cjlin/liblinear/. As for the hash function, we experimented with both the hashCode of the Java String class, and murmurHash. We found that the results were not significantly different from one another in terms of the number of hash collisions and classification accuracy. We also experimented with both $\xi :\mathcal{S}\to \left\{\pm 1\right\}$ and ξ ≡ 1  actual counts, and found that the results were not significantly different. The results shown in the next subsection use the hashCode function and ξ ≡ 1. On all three data sets, we reported the average classification accuracy obtained in a 5fold cross validation experiment. The results are statistically significant (p < 0.05). The classification accuracy is shown as a function of the number of features. The x axis of all figures in the next subsection shows the number of features on a log_{2} scale (i.e., number of bits in the hashtable).
Results
Comparison of fixed length with variable length kgram representations
Comparison of fixedlength with variablelength kgram representations.
Bag of fixed or variable length kgrams  nonplant  

Accuracy %  # features  
1grams  71.21  20 
2grams  70.85  400 
3grams  79.80  7999 
4grams  79.03  146598 
(12)grams  70.56  420 
(13)grams  79.69  8419 
(14)grams  82.83  155017 
(15)grams  80.09  950849 
The number of variable length kgrams, for k ranging from 1 to 4, is 155,017. Feature hashing eliminates the need for storing the vocabularies in memory by implicitly encoding the mapping from strings to integers into a hash function. We conclude that feature hashing is very effective on prohibitively highdimensional kgram representations, which would otherwise be impractical to use. Because (14)gram representation results in the highest performance, we used it for subsequent experiments.
The influence of hash sizes on classifiers' performance and the comparison of feature hashing with baseline (i.e., the "bag of kgrams" approach)
The number of variablelength kgrams and the rate of hash collisions for various hash sizes.
Value of b  nonplant  plant  psortNeg  

# features  Collisions %  # features  Collisions %  # features  Collisions %  
2^{22}  155017  0  111544  0  124389  0 
2^{20}  153166  1.21  110236  1.18  122894  1.22 
2^{19}  147223  5.29  107299  3.95  118871  4.64 
2^{18}  132754  16.30  99913  11.43  109535  13.22 
2^{17}  99764  45.04  82141  31.38  87618  35.66 
2^{16}  59358  78.53  53616  64.29  55555  68.85 
2^{15}  32474  95.80  31788  89.56  32075  92.02 
2^{14}  16384  100  16384  100  16384  100 
As the hash size increases beyond 2^{16}, the performance of SVM classifiers does not change substantially, and, eventually, converges. For example, on the nonplant data set, with 2^{16} hash size, SVM achieves 81.3% accuracy, whereas with 2^{22} hash size, SVM achieves an accuracy of 82.83% (Figure 4a). On the plant data set, SVMs achieve 78.4% and 78.51% accuracy, with 2^{16} and 2^{22} hash sizes, respectively (Figure 4b). Furthermore, as the hash size increases beyond 2^{16}, the percentage of hash collisions decreases until no collisions occur (Table 2). For all three data sets, with 2^{22} hash size, there are no hash collisions. The performance of SVMs trained on hash features in the 2^{22} dimensional space is matched by that of SVMs trained on hash features in the 2^{18} dimensional space, suggesting that the hash collisions beyond 2^{18} does not significantly distort the data.
Because 2^{22} (= 4,194,304) highly exceeds the number of unique features, and the rate of hash collisions becomes zero, this can be regarded as equivalent to the classifiers trained without hashing, which require storing the vocabularies in memory, referred as baseline (or the "bag of kgrams") (Figure 4). Moreover, we considered 2^{16} as the point where the performance starts degrading. Note that the vocabulary sizes, i.e., the number of unique variable length kgrams, for nonplant, plant, and psortNeg, are 155017, 111544, and 124389, respectively.
We conclude that, if feature hashing is used to reduce dimensionality from very large dimensions, e.g., 2^{22} to midsize dimensions, e.g., 2^{16}, the hash collisions do not substantially hurt the classification accuracy, whereas if it is used to reduce dimensionality from midsize dimensions to smaller dimensions, e.g., 2^{10}, the hash collision significantly distort the data, and the corresponding SVMs result in poor performance. Also, feature hashing makes it possible to train SVMs that use substantially smaller number of dimensions compared to the baseline, for a small or no drop in accuracy, for example, for a hash size of 2^{16} = 65536 (compared to 155017 variablelength kgrams on the nonplant data set).
Conclusion
We presented an application of feature hashing to reduce dimensionality of very highdimensional feature vectors to midsize feature vectors on protein sequence data and compared it with the "bag of kgrams" approach.
The results of our experiments on three protein subcellular localization data sets show that feature hashing is an effective approach to dealing with prohibitively highdimensional variable length kgram representations. Feature hashing makes it possible to train SVM classifiers that use substantially smaller number of features compared to the approach which requires storing the vocabularies in memory, i.e., the "bag of kgrams" approach, while resulting in a small or no decrease in classification performance.
Because recent advances in sequencing technologies have resulted in an exponential increase in the rate at which DNA and protein sequence data are being acquired [28], the application of feature hashing on biological sequence data advances the current state of the art in terms of algorithms that can efficiently process highdimensional data into lowdimensional feature vectors at runtime.
In the future, it would be interesting to investigate how the performance of hash kernels compares to that of histogrambased motif kernels for protein sequences, introduced by Ong and Zien [29], and the mismatch string kernels for SVM protein classification introduced by Lesli et al. [30]. Along the lines of dimensionality reduction, it would be interesting to compare the performance of feature hashing with that of feature abstraction [17] on protein sequence classification tasks. Furthermore, another direction is to apply feature hashing to other types of biological sequence data, e.g., DNA data, and other tasks, e.g., protein function prediction.
Declarations
Acknowledgements
This article has been published as part of Proteome Science Volume 10 Supplement 1, 2012: Selected articles from the IEEE International Conference on Bioinformatics and Biomedicine 2011: Proteome Science. The full contents of the supplement are available online at http://www.proteomesci.com/supplements/10/S1.
This research was funded in part by an NSF grant #0845487 to Prasenjit Mitra.
Authors’ Affiliations
References
 Baldi P, Brunak S: Bioinformatics: the Machine Learning Approach. MIT Press; 2001.Google Scholar
 Jolliffe IT: Principal Component Analysis. SpringerVerlag, New York; 1986.View ArticleGoogle Scholar
 Blei D, Ng A, Jordan M: Latent Dirichlet Allocation. Journal of Machine Learning Research 2003, 3: 993–1022.Google Scholar
 Hofmann T: Probabilistic Latent Semantic Analysis. Proc of UAI'99, Stockholm 1999.Google Scholar
 Guyon I, Elisseeff A: An introduction to variable and feature selection. J Mach Learn Res 2003, 3: 1157–1182.Google Scholar
 Fleuret F: Fast Binary Feature Selection with Conditional Mutual Information. J Mach Learn Res 2004, 5: 1531–1555.Google Scholar
 Yang Y, Pederson JO: Feature selection in statistical learning of text categorization. In Proceedings of ICML 1997, 412–420.Google Scholar
 Shi Q, Petterson J, Dror G, Langford J, Smola A, Vishwanathan S: Hash Kernels for Structured Data. J Mach Learn Res 2009, 10: 2615–2637.Google Scholar
 Weinberger K, Dasgupta A, Attenberg J, Langford J, Smola A: Feature Hashing for Large Scale Multitask Learning. Proc of ICML 2009.Google Scholar
 Forman G, Kirshenbaum E: Extremely fast text feature extraction for classification and indexing. Proc of the 17th ACM CIKM, New York, NY, USA 2008, 1221–1230.Google Scholar
 Langford J, Li L, Strehl A: Vowpal Wabbit online learning project. 2007.Google Scholar
 Kira K, Rendell LA: The feature selection problem: Traditional methods and a new algorithm. Proc of AAAI, San Jose, CA 1992, 122–134.Google Scholar
 Saeys Y, Inza IN, Larrañaga P: A review of feature selection techniques in bioinformatics. Bioinformatics 2007, 23: 2507–2517. 10.1093/bioinformatics/btm344PubMedView ArticleGoogle Scholar
 Deerwester S, Dumais ST, Furnas GW, Landauer TK, Harshman R: Indexing by latent semantic analysis. Journal of the American Society for Information Science 1990,41(6):391–407. 10.1002/(SICI)10974571(199009)41:6<391::AIDASI1>3.0.CO;29View ArticleGoogle Scholar
 Airoldi E, Blei D, Fienberg S, Xing E: Mixed membership stochastic block models for relational data with application to proteinprotein interactions. In Proc of the Intl Biometrics SocietyENAR Annual Meeting 2006.Google Scholar
 Pan XY, Zhang YN, Shen HB: LargeScale Prediction of Human ProteinProtein Interactions from Amino Acid Sequence Based on Latent Topic Features. Journal of Proteome Research 2010,9(10):4992–5001. 10.1021/pr100618tPubMedView ArticleGoogle Scholar
 Silvescu A, Caragea C, Honavar V: Combining SuperStructuring and Abstraction on Sequence Classification. ICDM 2009, 986–991.Google Scholar
 Ganchev K, Dredze M: Small Statistical Models by Random Feature Mixing. Proceedings of the ACL2008 Workshop on Mobile Language Processing, Association for Computational Linguistics 2008.Google Scholar
 Wesselink JJ, delaIglesia B, James SA, Dicks JL, Roberts IN, RaywardSmith V: Determining a unique defining DNA sequence for yeast species using hashing techniques. Bioinformatics 2002,18(2):1004–10.PubMedView ArticleGoogle Scholar
 Buhler J, Tompa M: Finding motifs using random projections. Proceedings of the Fifth Annual International Conference on Computational Biology, RECOMB '01 2001, 69–76.View ArticleGoogle Scholar
 Indyk P, Motwani R: Approximate nearest neighbors: towards removing the curse of dimensionality. Proceedings of the Thirtieth Annual ACM Symposium on Theory of Computing, STOC '98 1998, 604–613.View ArticleGoogle Scholar
 Buhler J: Efficient LargeScale Sequence Comparison by LocalitySensitive Hashing. Bioinformatics 2001,17(5):419–428. 10.1093/bioinformatics/17.5.419PubMedView ArticleGoogle Scholar
 Durbin R, Eddy SR, Krogh A, Mitchison G: Biological sequence analysis: Probabilistic Models of Proteins and Nucleic Acids. Cambridge University Press; 2004.Google Scholar
 Begleiter R, ElYaniv R, Yona G: On Prediction Using Variable Order Markov Models. Journal of Artificial Intelligence Res 2004, 22: 385–421.Google Scholar
 Gardy JL, Spencer C, Wang K, Ester M, Tusnady GE, Simon I, Hua S, deFays K, Lambert C, Nakai K, Brinkman FS: PSORTB: improving protein subcellular localization prediction for Gramnegative bacteria. Nucleic Acids Research 2003,31(13):3613–17. 10.1093/nar/gkg602PubMed CentralPubMedView ArticleGoogle Scholar
 Emanuelsson O, Nielsen H, Brunak S, von Heijne G: Predicting subcellular localization of proteins based on their Nterminal amino acid sequence. J Mol Biol 2000, 300: 1005–1016. 10.1006/jmbi.2000.3903PubMedView ArticleGoogle Scholar
 Fan RE, Chang KW, Hsieh CJ, Wang XR, Lin CJ: LIBLINEAR: A library for large linear classification. J of Machine Learning Res 2008, 9: 1871–1874.Google Scholar
 Ansorge W: Nextgeneration DNA sequencing techniques. New Biotechnology 2009,25(4):195–203. 10.1016/j.nbt.2008.12.009PubMedView ArticleGoogle Scholar
 Ong CS, Zien A: An Automated Combination of Kernels for Predicting Protein Subcellular Localization. Proc of WABI 2008, 186–179.Google Scholar
 Leslie C, Eskin E, Weston J, Noble WS: Mismatch String Kernels for SVM Protein Classification. Advances in Neural Information Processing Systems (NIPS 2002) 2002.Google Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.