Fast protein homology and fold detection with sparse spatial sample kernels

Kuksa, Pavel P.; Huang, Pai-Hsi; Pavlović, Vladimir

doi:10.1109/icpr.2008.4761450

Cited by 11 publications

(15 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Baseline +Local ASK Spectrum (n-gram) [18] 27.91 33.06 Mismatch [19] 41.92 46.68 Spatial sample kernel [15] 50.12 52.75 Semi-supervised Cluster kernel [28] 67.91 70.14 Furthermore the cluster kernel introduces new examples (sequences) and requires semi-supervision at testing time, while our unsupervised auxiliary tasks are feature learning methods, i.e. the learned features could be directly added to the existing feature set.…”

Section: Methodsmentioning

confidence: 99%

“…For this problem, we use a popular benchmark dataset for structural homology prediction (SCOP) that corresponds to 54 remote homology detection experiments [28,17]. We test local ASK (with local embedding trained on a UNIPROT dataset, a collection of about 400,000 protein sequences) and compare with the supervised string kernels commonly used for the remote homology detection [19,28,15,17]. Each amino acid is treated as a word in this case.…”

Section: Task 4: Comparison On Biological Sequence Taskmentioning

confidence: 99%

See 1 more Smart Citation

Semi-supervised Abstraction-Augmented String Kernel for Multi-level Bio-Relation Extraction

Kuksa

Bai

et al. 2010

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. Bio-relation extraction (bRE), an important goal in bio-text mining, involves subtasks identifying relationships between bio-entities in text at multiple levels, e.g., at the article, sentence or relation level. A key limitation of current bRE systems is that they are restricted by the availability of annotated corpora. In this work we introduce a semisupervised approach that can tackle multi-level bRE via string comparisons with mismatches in the string kernel framework. Our string kernel implements an abstraction step, which groups similar words to generate more abstract entities, which can be learnt with unlabeled data. Specifically, two unsupervised models are proposed to capture contextual (local or global) semantic similarities between words from a large unannotated corpus. This Abstraction-augmented String Kernel (ASK) allows for better generalization of patterns learned from annotated data and provides a unified framework for solving bRE with multiple degrees of detail. ASK shows effective improvements over classic string kernels on four datasets and achieves state-of-the-art bRE performance without the need for complex linguistic features.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Task 4: Comparison On Biological Sequence Taskmentioning

confidence: 99%

Semi-supervised Abstraction-Augmented String Kernel for Multi-level Bio-Relation Extraction

Kuksa

Bai

et al. 2010

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…Weston et al in [11] and Kuksa et al in [7] show that the discriminative power of the classifiers improve significantly using this neighborhood information. However, as we show in the subsequent sections, the way N (X) is constructed impacts in a large way the accuracy of the neighborhood methods.…”

Section: Introductionmentioning

confidence: 98%

“…Traditional mismatch kernels do not take into account any spatial information contained in k-mers which may be critical for accurate modeling of inexact relationships. More recently, Kuksa et al introduced the sparse spatial sample kernels (SSSK) [7] which model the intrinsic spatial information in substrings of X. In particular, they…”

Section: Introductionmentioning

confidence: 99%

“…The unlabeled sequences serve as conduits for propagating information between distant homologs. The families of profile [6] kernels, the sequence neighborhood kernels [11] and the SSSK [7] achieve stateof-the-art performance by leveraging this source of information. All of the methods rely on the notion of sequence neighborhoods N (X), the sets of sequences most similar to any particular sequence X. N (X) is typically established using a scoring function s(X, X ), such as the evalue, yielding N (X) = {X : s(X, X ) ≤ δ} for a fixed threshold δ.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

On the Role of Local Matching for Efficient Semi-supervised Protein Sequence Classification

Kuksa

Huang

Pavlović

2008

2008 IEEE International Conference on Bioinformatics and Biomedicine

Self Cite

View full text Add to dashboard Cite

Recent studies in protein sequence analysis have leveraged the power of unlabeled data. For example, the profile and mismatch neighborhood kernels have shown significant improvements over classifiers estimated under the fully supervised setting. In this study, we present a principled and biologically motivated framework that more effectively exploits the unlabeled data by only utilizing regions that are more likely to be biologically relevant for better prediction accuracy. As overly-represented sequences in large uncurated databases may bias kernel estimations that rely on unlabeled data, we also propose a method to remove this bias and improve performance of resulting classifiers. Combined with a computationally efficient sparse family of string kernels, our proposed framework achieves state-ofthe-art accuracy in semi-supervised protein remote homology detection on three large unlabeled databases.

show abstract

DNA Methylation Prediction Using Reduced Features Obtained via Gappy Pair Kernel and Partial Least Square

Shah

Rahman

Jabeen³

et al. 2022

IEEE Access

View full text Add to dashboard Cite

Fast protein homology and fold detection with sparse spatial sample kernels

Cited by 11 publications

References 10 publications

Semi-supervised Abstraction-Augmented String Kernel for Multi-level Bio-Relation Extraction

Semi-supervised Abstraction-Augmented String Kernel for Multi-level Bio-Relation Extraction

On the Role of Local Matching for Efficient Semi-supervised Protein Sequence Classification

DNA Methylation Prediction Using Reduced Features Obtained via Gappy Pair Kernel and Partial Least Square

Contact Info

Product

Resources

About