Protein sequence classification using feature hashing

Recent advances in next-generation sequencing technologies have resulted in an exponential increase in the rate at which protein sequence data are being acquired. The k-gram 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 high-dimensional space is "reduced" by hashing the features into a low-dimensional 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 k-grams" approach. Our results show that feature hashing is an effective approach to reducing dimensionality on protein sequence classification tasks.     

Classifying genomic sequences by sequence feature analysis

Traditional sequence analysis depends on sequence alignment. In this study, we analyzed various functional regions of the human genome based on sequence features, including word frequency, dinucleotide relative abundance, and base-base correlation. We analyzed the human chromosome 22 and classified the upstream, exon, intron, downstream, and intergenic regions by principal component analysis and discriminant analysis of these features. The results show that we could classify the functional regions of genome based on sequence feature and discriminant analysis.     

Mesoscale DNA feature in antibody-coding sequence facilitates somatic hypermutation

1. Download : Download high-res image (163KB)
2. Download : Download full-size image

     

基于序列特征网络的蛋白质结构类型研究

利用复杂网络的方法来探索序列特征因素对蛋白质结构的影响。由于蛋白质的序列对结构具有重要且复杂的影响,因此将蛋白质的结构以及序列特征之间的关系模拟成一个复杂系统,通过利用互相关系数、标准化互信息和传递熵等方法来建立以序列特征为节点的加权网络,进而利用网络中心性的方法来分析不同蛋白质结构类型对应加权网络的中心性分布的差异,探索不同结构类型蛋白质的序列特征差异。发现不同的蛋白质结构类型对应的序列特征网络既有共性又有差异,文章将针对每一种结构类型的网络中心性分布,以及不同结构类型之间的共性与差异进行详细地讨论。研究结果对蛋白质序列与结构之间关系的研究,特别是结构分类研究具有重要的意义。     

Predicting disulfide connectivity from protein sequence using multiple sequence feature vectors and secondary structure

MOTIVATION: Disulfide bonds are primary covalent crosslinks between two cysteine residues in proteins that play critical roles in stabilizing the protein structures and are commonly found in extracy-toplasmatic or secreted proteins. In protein folding prediction, the localization of disulfide bonds can greatly reduce the search in conformational space. Therefore, there is a great need to develop computational methods capable of accurately predicting disulfide connectivity patterns in proteins that could have potentially important applications. RESULTS: We have developed a novel method to predict disulfide connectivity patterns from protein primary sequence, using a support vector regression (SVR) approach based on multiple sequence feature vectors and predicted secondary structure by the PSIPRED program. The results indicate that our method could achieve a prediction accuracy of 74.4% and 77.9%, respectively, when averaged on proteins with two to five disulfide bridges using 4-fold cross-validation, measured on the protein and cysteine pair on a well-defined non-homologous dataset. We assessed the effects of different sequence encoding schemes on the prediction performance of disulfide connectivity. It has been shown that the sequence encoding scheme based on multiple sequence feature vectors coupled with predicted secondary structure can significantly improve the prediction accuracy, thus enabling our method to outperform most of other currently available predictors. Our work provides a complementary approach to the current algorithms that should be useful in computationally assigning disulfide connectivity patterns and helps in the annotation of protein sequences generated by large-scale whole-genome projects. AVAILABILITY: The prediction web server and Supplementary Material are accessible at http://foo.maths.uq.edu.au/~huber/disulfide     

Searching sequence space for high-affinity binding peptides using ribosome display

We present the construction of a synthetic library based on the scaffold of bovine heart fatty acid-binding protein (FABP) with 1.1x10(14) independent members. Ribosome display was applied to select streptavidin-binding peptides in vitro from 2x10(13) molecules of the library each encoding FABP with 15 contiguous random amino acid residues at its N terminus. The selection yielded several different binding peptides. The best binder possessed a dissociation constant as low as 4nM and, in contrast to the previously isolated peptides, contained no HPQ motif. A substitution analysis enabled shortening of the 15-mer peptide and revealed a 9-mer variant with a dissociation constant of 17nM, which is a 1000-fold increase of affinity compared to the already known peptides of this size. This high-affinity binding peptide in combination with the whole set of streptavidin conjugates should be an extremely useful tool for the detection and purification of recombinant proteins.     

An artificial intelligence approach to DNA sequence feature recognition.

The ultimate goal of the Human Genome project is to extract the biologically relevant information recorded in the estimated 100,000 genes encoded by the 3 x 10(9) bases of the human genome. This necessitates development of reliable computer-based methods capable of analysing and correctly identifying genes in the vast amounts of DNA-sequence data generated. Such tools may save time and labour by simplifying, for example, screening of cDNA libraries. They may also facilitate the localization of human disease genes by identifying candidate genes in promising regions of anonymous DNA sequence.     

SEAN: SNP prediction and display program utilizing EST sequence clusters

SEAN is an application that predicts single nucleotide polymorphisms (SNPs) using multiple sequence alignments produced from expressed sequence tag (EST) clusters. The algorithm uses rules of sequence identity and SNP abundance to determine the quality of the prediction. A Java viewer is provided to display the EST alignments and predicted SNPs.     

Quantitative assessment of peptide sequence diversity in M13 combinatorial peptide phage display libraries

Novel statistical methods have been developed and used to quantitate and annotate the sequence diversity within combinatorial peptide libraries on the basis of small numbers (1-200) of sequences selected at random from commercially available M13 p3-based phage display libraries. These libraries behave statistically as though they correspond to populations containing roughly 4.0+/-1.6% of the random dodecapeptides and 7.9+/-2.6% of the random constrained heptapeptides that are theoretically possible within the phage populations. Analysis of amino acid residue occurrence patterns shows no demonstrable influence on sequence censorship by Escherichia coli tRNA isoacceptor profiles or either overall codon or Class II codon usage patterns, suggesting no metabolic constraints on recombinant p3 synthesis. There is an overall depression in the occurrence of cysteine, arginine and glycine residues and an overabundance of proline, threonine and histidine residues. The majority of position-dependent amino acid sequence bias is clustered at three positions within the inserted peptides of the dodecapeptide library, +1, +3 and +12 downstream from the signal peptidase cleavage site. Conformational tendency measures of the peptides indicate a significant preference for inserts favoring a beta-turn conformation. The observed protein sequence limitations can primarily be attributed to genetic codon degeneracy and signal peptidase cleavage preferences. These data suggest that for applications in which maximal sequence diversity is essential, such as epitope mapping or novel receptor identification, combinatorial peptide libraries should be constructed using codon-corrected trinucleotide cassettes within vector-host systems designed to minimize morphogenesis-related censorship.     

Prediction of protein subcellular location using a combined feature of sequence

To understand the structure and function of a protein, an important task is to know where it occurs in the cell. Thus, a computational method for properly predicting the subcellular location of proteins would be significant in interpreting the original data produced by the large-scale genome sequencing projects. The present work tries to explore an effective method for extracting features from protein primary sequence and find a novel measurement of similarity among proteins for classifying a protein to its proper subcellular location. We considered four locations in eukaryotic cells and three locations in prokaryotic cells, which have been investigated by several groups in the past. A combined feature of primary sequence defined as a 430D (dimensional) vector was utilized to represent a protein, including 20 amino acid compositions, 400 dipeptide compositions and 10 physicochemical properties. To evaluate the prediction performance of this encoding scheme, a jackknife test based on nearest neighbor algorithm was employed. The prediction accuracies for cytoplasmic, extracellular, mitochondrial, and nuclear proteins in the former dataset were 86.3%, 89.2%, 73.5% and 89.4%, respectively, and the total prediction accuracy reached 86.3%. As for the prediction accuracies of cytoplasmic, extracellular, and periplasmic proteins in the latter dataset, the prediction accuracies were 97.4%, 86.0%, and 79.7, respectively, and the total prediction accuracy of 92.5% was achieved. The results indicate that this method outperforms some existing approaches based on amino acid composition or amino acid composition and dipeptide composition.     

Inference of global HIV-1 sequence patterns and preliminary feature analysis

The epidemiology of HIV-1 varies in different areas of the world, and it is possible that this complexity may leave unique footprints in the viral genome. Thus, we attempted to find significant patterns in global HIV-1 genome sequences. By applying the rule inference algorithm RIPPER (Repeated Incremental Pruning to Produce Error Reduction) to multiple sequence alignments of Env sequences from four classes of compiled datasets, we generated four sets of signature patterns. We found that these patterns were able to distinguish southeastern Asian from nonsoutheastern Asian sequences with 97.5% accuracy, Chinese from non-Chinese sequences with 98.3% accuracy, African from non-African sequences with 88.4% accuracy, and southern African from non-southern African sequences with 91.2% accuracy. These patterns showed different associations with subtypes and with amino acid positions. In addition, some signature patterns were characteristic of the geographic area from which the sample was taken. Amino acid features corresponding to the phylogenetic clustering of HIV-1 sequences were consistent with some of the deduced patterns. Using a combination of patterns inferred from subtypes B, C, and all subtypes chimeric with CRF01_AE worldwide, we found that signature patterns of subtype C were extremely common in some sampled countries (for example, Zambia in southern Africa), which may hint at the origin of this HIV-1 subtype and the need to pay special attention to this area of Africa. Signature patterns of subtype B sequences were associated with different countries. Even more, there are distinct patterns at single position 21 with glycine, leucine and isoleucine corresponding to subtype C, B and all possible recombination forms chimeric with CRF01_AE, which also indicate distinct geographic features. Our method widens the scope of inference of signature from geographic, genetic, and genomic viewpoints. These findings may provide a valuable reference for epidemiological research or vaccine design.     

Using phage display to select antibodies recognizing post-translational modifications independently of sequence context

Many cellular activities are controlled by post-translational modifications, the study of which is hampered by the lack of specific reagents due in large part to their ubiquitous and non-immunogenic nature. Although antibodies against specifically modified sequences are relatively easy to obtain, it is extremely difficult to derive reagents recognizing post-translational modifications independently of the sequence context surrounding the modification. In this study, we examined the possibility of selecting such antibodies from large phage antibody libraries using sulfotyrosine as a test case. Sulfotyrosine is a post-translational modification important in many extracellular protein-protein interactions, including human immunodeficiency virus infection. After screening almost 8000 selected clones, we were able to isolate a single specific single chain Fv using two different selection strategies, one of which included elution with tyrosine sulfate. This antibody was able to recognize sulfotyrosine independently of its sequence context in test peptides and a number of different natural proteins. Antibody reactivity was lost by antigen treatment with sulfatase or preincubation with soluble tyrosine sulfate, indicating its specificity. The isolation of this antibody signals the potential of phage antibody libraries in the derivation of reagents specific for post-translational modifications, although the extensive screening required indicates that such antibodies are extremely rare.