Sparse representation approaches for the classification of high-dimensional biological data

Background

High-throughput genomic and proteomic data have important applications in medicine including prevention, diagnosis, treatment, and prognosis of diseases, and molecular biology, for example pathway identification. Many of such applications can be formulated to classification and dimension reduction problems in machine learning. There are computationally challenging issues with regards to accurately classifying such data, and which due to dimensionality, noise and redundancy, to name a few. The principle of sparse representation has been applied to analyzing high-dimensional biological data within the frameworks of clustering, classification, and dimension reduction approaches. However, the existing sparse representation methods are inefficient. The kernel extensions are not well addressed either. Moreover, the sparse representation techniques have not been comprehensively studied yet in bioinformatics.

Results

In this paper, a Bayesian treatment is presented on sparse representations. Various sparse coding and dictionary learning models are discussed. We propose fast parallel active-set optimization algorithm for each model. Kernel versions are devised based on their dimension-free property. These models are applied for classifying high-dimensional biological data.

Conclusions

In our experiment, we compared our models with other methods on both accuracy and computing time. It is shown that our models can achieve satisfactory accuracy, and their performance are very efficient.

     

Effective norm emergence in cell systems under limited communication

Background

The cooperation of cells in biological systems is similar to that of agents in cooperative multi-agent systems. Research findings in multi-agent systems literature can provide valuable inspirations to biological research. The well-coordinated states in cell systems can be viewed as desirable social norms in cooperative multi-agent systems. One important research question is how a norm can rapidly emerge with limited communication resources.

Results

In this work, we propose a learning approach which can trade off the agents’ performance of coordinating on a consistent norm and the communication cost involved. During the learning process, the agents can dynamically adjust their coordination set according to their own observations and pick out the most crucial agents to coordinate with. In this way, our method significantly reduces the coordination dependence among agents.

Conclusion

The experiment results show that our method can efficiently facilitate the social norm emergence among agents, and also scale well to large-scale populations.

     

Deep learning for DNase I hypersensitive sites identification

Background

The DNase I hypersensitive sites (DHSs) are associated with the cis-regulatory DNA elements. An efficient method of identifying DHSs can enhance the understanding on the accessibility of chromatin. Despite a multitude of resources available on line including experimental datasets and computational tools, the complex language of DHSs remains incompletely understood.

Methods

Here, we address this challenge using an approach based on a state-of-the-art machine learning method. We present a novel convolutional neural network (CNN) which combined Inception like networks with a gating mechanism for the response of multiple patterns and longterm association in DNA sequences to predict multi-scale DHSs in Arabidopsis, rice and Homo sapiens.

Results

Our method obtains 0.961 area under curve (AUC) on Arabidopsis, 0.969 AUC on rice and 0.918 AUC on Homo sapiens.

Conclusions

Our method provides an efficient and accurate way to identify multi-scale DHSs sequences by deep learning.

     

Intrinsic disorder in putative protein sequences

Background

Intrinsically disordered proteins (IDPs) and regions (IDRs) perform a variety of crucial biological functions despite lacking stable tertiary structure under physiological conditions in vitro. State-of-the-art sequence-based predictors of intrinsic disorder are achieving per-residue accuracies over 80%. In a genome-wide study of intrinsic disorder in human genome we observed a big difference in predicted disorder content between confirmed and putative human proteins. We investigated a hypothesis that this discrepancy is not correct, and that it is due to incorrectly annotated parts of the putative protein sequences that exhibit some similarities to confirmed IDRs, which lead to high predicted disorder content.

Methods

To test this hypothesis we trained a predictor to discriminate sequences of real proteins from synthetic sequences that mimic errors of gene finding algorithms. We developed a procedure to create synthetic peptide sequences by translation of non-coding regions of genomic sequences and translation of coding regions with incorrect codon alignment.

Results

Application of the developed predictor to putative human protein sequences showed that they contain a substantial fraction of incorrectly assigned regions. These regions are predicted to have higher levels of disorder content than correctly assigned regions. This partially, albeit not completely, explains the observed discrepancy in predicted disorder content between confirmed and putative human proteins.

Conclusions

Our findings provide the first evidence that current practice of predicting disorder content in putative sequences should be reconsidered, as such estimates may be biased.

     

Predicting enhancers with deep convolutional neural networks

Background

With the rapid development of deep sequencing techniques in the recent years, enhancers have been systematically identified in such projects as FANTOM and ENCODE, forming genome-wide landscapes in a series of human cell lines. Nevertheless, experimental approaches are still costly and time consuming for large scale identification of enhancers across a variety of tissues under different disease status, making computational identification of enhancers indispensable.

Results

To facilitate the identification of enhancers, we propose a computational framework, named DeepEnhancer, to distinguish enhancers from background genomic sequences. Our method purely relies on DNA sequences to predict enhancers in an end-to-end manner by using a deep convolutional neural network (CNN). We train our deep learning model on permissive enhancers and then adopt a transfer learning strategy to fine-tune the model on enhancers specific to a cell line. Results demonstrate the effectiveness and efficiency of our method in the classification of enhancers against random sequences, exhibiting advantages of deep learning over traditional sequence-based classifiers. We then construct a variety of neural networks with different architectures and show the usefulness of such techniques as max-pooling and batch normalization in our method. To gain the interpretability of our approach, we further visualize convolutional kernels as sequence logos and successfully identify similar motifs in the JASPAR database.

Conclusions

DeepEnhancer enables the identification of novel enhancers using only DNA sequences via a highly accurate deep learning model. The proposed computational framework can also be applied to similar problems, thereby prompting the use of machine learning methods in life sciences.

     

Using multitask classification methods to investigate the kinase-specific phosphorylation sites

Background

Identification of phosphorylation sites by computational methods is becoming increasingly important because it reduces labor-intensive and costly experiments and can improve our understanding of the common properties and underlying mechanisms of protein phosphorylation.

Methods

A multitask learning framework for learning four kinase families simultaneously, instead of studying each kinase family of phosphorylation sites separately, is presented in the study. The framework includes two multitask classification methods: the Multi-Task Least Squares Support Vector Machines (MTLS-SVMs) and the Multi-Task Feature Selection (MT-Feat3).

Results

Using the multitask learning framework, we successfully identify 18 common features shared by four kinase families of phosphorylation sites. The reliability of selected features is demonstrated by the consistent performance in two multi-task learning methods.

Conclusions

The selected features can be used to build efficient multitask classifiers with good performance, suggesting they are important to protein phosphorylation across 4 kinase families.

     

Kernel-PCA data integration with enhanced interpretability

Background

Nowadays, combining the different sources of information to improve the biological knowledge available is a challenge in bioinformatics. One of the most powerful methods for integrating heterogeneous data types are kernel-based methods. Kernel-based data integration approaches consist of two basic steps: firstly the right kernel is chosen for each data set; secondly the kernels from the different data sources are combined to give a complete representation of the available data for a given statistical task.

Results

We analyze the integration of data from several sources of information using kernel PCA, from the point of view of reducing dimensionality. Moreover, we improve the interpretability of kernel PCA by adding to the plot the representation of the input variables that belong to any dataset. In particular, for each input variable or linear combination of input variables, we can represent the direction of maximum growth locally, which allows us to identify those samples with higher/lower values of the variables analyzed.

Conclusions

The integration of different datasets and the simultaneous representation of samples and variables together give us a better understanding of biological knowledge.

     

Predicting adverse drug reactions through interpretable deep learning framework

Background

Adverse drug reactions (ADRs) are unintended and harmful reactions caused by normal uses of drugs. Predicting and preventing ADRs in the early stage of the drug development pipeline can help to enhance drug safety and reduce financial costs.

Methods

In this paper, we developed machine learning models including a deep learning framework which can simultaneously predict ADRs and identify the molecular substructures associated with those ADRs without defining the substructures a-priori.

Results

We evaluated the performance of our model with ten different state-of-the-art fingerprint models and found that neural fingerprints from the deep learning model outperformed all other methods in predicting ADRs. Via feature analysis on drug structures, we identified important molecular substructures that are associated with specific ADRs and assessed their associations via statistical analysis.

Conclusions

The deep learning model with feature analysis, substructure identification, and statistical assessment provides a promising solution for identifying risky components within molecular structures and can potentially help to improve drug safety evaluation.

     

Extracting DNA words based on the sequence features: non-uniform distribution and integrity

Background

DNA sequence can be viewed as an unknown language with words as its functional units. Given that most sequence alignment algorithms such as the motif discovery algorithms depend on the quality of background information about sequences, it is necessary to develop an ab initio algorithm for extracting the “words” based only on the DNA sequences.

Methods

We considered that non-uniform distribution and integrity were two important features of a word, based on which we developed an ab initio algorithm to extract “DNA words” that have potential functional meaning. A Kolmogorov-Smirnov test was used for consistency test of uniform distribution of DNA sequences, and the integrity was judged by the sequence and position alignment. Two random base sequences were adopted as negative control, and an English book was used as positive control to verify our algorithm. We applied our algorithm to the genomes of Saccharomyces cerevisiae and 10 strains of Escherichia coli to show the utility of the methods.

Results

The results provide strong evidences that the algorithm is a promising tool for ab initio building a DNA dictionary.

Conclusions

Our method provides a fast way for large scale screening of important DNA elements and offers potential insights into the understanding of a genome.

     

HAlign-II: efficient ultra-large multiple sequence alignment and phylogenetic tree reconstruction with distributed and parallel computing

Background

Multiple sequence alignment (MSA) plays a key role in biological sequence analyses, especially in phylogenetic tree construction. Extreme increase in next-generation sequencing results in shortage of efficient ultra-large biological sequence alignment approaches for coping with different sequence types.

Methods

Distributed and parallel computing represents a crucial technique for accelerating ultra-large (e.g. files more than 1 GB) sequence analyses. Based on HAlign and Spark distributed computing system, we implement a highly cost-efficient and time-efficient HAlign-II tool to address ultra-large multiple biological sequence alignment and phylogenetic tree construction.

Results

The experiments in the DNA and protein large scale data sets, which are more than 1GB files, showed that HAlign II could save time and space. It outperformed the current software tools. HAlign-II can efficiently carry out MSA and construct phylogenetic trees with ultra-large numbers of biological sequences. HAlign-II shows extremely high memory efficiency and scales well with increases in computing resource.

Conclusions

THAlign-II provides a user-friendly web server based on our distributed computing infrastructure. HAlign-II with open-source codes and datasets was established at http://lab.malab.cn/soft/halign.

     

A multi-context learning approach for EEG epileptic seizure detection

Background

Epilepsy is a neurological disease characterized by unprovoked seizures in the brain. The recent advances in sensor technologies allow researchers to analyze the collected biological records to improve the treatment of epilepsy. Electroencephalogram (EEG) is the most commonly used biological measurement to effectively capture the abnormalities of different brain areas during the EEG seizures. To avoid manual visual inspection from long-term EEG readings, automatic epileptic EEG seizure detection has become an important research issue in bioinformatics.

Results

We present a multi-context learning approach to automatically detect EEG seizures by incorporating a feature fusion strategy. We generate EEG scalogram sequences from the EEG records by utilizing waveform transform to describe the frequency content over time. We propose a multi-stage unsupervised model that integrates the features extracted from the global handcrafted engineering, channel-wise deep learning, and EEG embeddings, respectively. The learned multi-context features are subsequently merged to train a seizure detector.

Conclusions

To validate the effectiveness of the proposed approach, extensive experiments against several baseline methods are carried out on two benchmark biological datasets. The experimental results demonstrate that the representative context features from multiple perspectives can be learned by the proposed model, and further improve the performance for the task of EEG seizure detection.

     

A stable iterative method for refining discriminative gene clusters

Background

Microarray technology is often used to identify the genes that are differentially expressed between two biological conditions. On the other hand, since microarray datasets contain a small number of samples and a large number of genes, it is usually desirable to identify small gene subsets with distinct pattern between sample classes. Such gene subsets are highly discriminative in phenotype classification because of their tightly coupling features. Unfortunately, such identified classifiers usually tend to have poor generalization properties on the test samples due to overfitting problem.

Results

We propose a novel approach combining both supervised learning with unsupervised learning techniques to generate increasingly discriminative gene clusters in an iterative manner. Our experiments on both simulated and real datasets show that our method can produce a series of robust gene clusters with good classification performance compared with existing approaches.

Conclusion

This backward approach for refining a series of highly discriminative gene clusters for classification purpose proves to be very consistent and stable when applied to various types of training samples.

     

Efficient mining gapped sequential patterns for motifs in biological sequences

Background

Pattern mining for biological sequences is an important problem in bioinformatics and computational biology. Biological data mining yield impact in diverse biological fields, such as discovery of co-occurring biosequences, which is important for biological data analyses. The approaches of mining sequential patterns can discover all-length motifs of biological sequences. Nevertheless, traditional approaches of mining sequential patterns inefficiently mine DNA and protein data since the data have fewer letters and lengthy sequences. Furthermore, gap constraints are important in computational biology since they cope with irrelative regions, which are not conserved in evolution of biological sequences.

Results

We devise an approach to efficiently mine sequential patterns (motifs) with gap constraints in biological sequences. The approach is the Depth-First Spelling algorithm for mining sequential patterns of biological sequences with Gap constraints (termed DFSG).

Conclusions

PrefixSpan is one of the most efficient methods in traditional approaches of mining sequential patterns, and it is the basis of GenPrefixSpan. GenPrefixSpan is an approach built on PrefixSpan with gap constraints, and therefore we compare DFSG with GenPrefixSpan. In the experimental results, DFSG mines biological sequences much faster than GenPrefixSpan.

     

Hot spot prediction in protein-protein interactions by an ensemble system

Background

Hot spot residues are functional sites in protein interaction interfaces. The identification of hot spot residues is time-consuming and laborious using experimental methods. In order to address the issue, many computational methods have been developed to predict hot spot residues. Moreover, most prediction methods are based on structural features, sequence characteristics, and/or other protein features.

Results

This paper proposed an ensemble learning method to predict hot spot residues that only uses sequence features and the relative accessible surface area of amino acid sequences. In this work, a novel feature selection technique was developed, an auto-correlation function combined with a sliding window technique was applied to obtain the characteristics of amino acid residues in protein sequence, and an ensemble classifier with SVM and KNN base classifiers was built to achieve the best classification performance.

Conclusion

The experimental results showed that our model yields the highest F1 score of 0.92 and an MCC value of 0.87 on ASEdb dataset. Compared with other machine learning methods, our model achieves a big improvement in hot spot prediction.

Availability

http://deeplearner.ahu.edu.cn/web/HotspotEL.htm.

     

Unsupervised Integration of Multiple Protein Disorder Predictors: The Method and Evaluation on CASP7, CASP8 and CASP9 Data

Background

Studies of intrinsically disordered proteins that lack a stable tertiary structure but still have important biological functions critically rely on computational methods that predict this property based on sequence information. Although a number of fairly successful models for prediction of protein disorder have been developed over the last decade, the quality of their predictions is limited by available cases of confirmed disorders.

Results

To more reliably estimate protein disorder from protein sequences, an iterative algorithm is proposed that integrates predictions of multiple disorder models without relying on any protein sequences with confirmed disorder annotation. The iterative method alternately provides the maximum a posterior (MAP) estimation of disorder prediction and the maximum-likelihood (ML) estimation of quality of multiple disorder predictors. Experiments on data used at CASP7, CASP8, and CASP9 have shown the effectiveness of the proposed algorithm.

Conclusions

The proposed algorithm can potentially be used to predict protein disorder and provide helpful suggestions on choosing suitable disorder predictors for unknown protein sequences.

     

Do universal codon-usage patterns minimize the effects of mutation and translation error?

Background

Do species use codons that reduce the impact of errors in translation or replication? The genetic code is arranged in a way that minimizes errors, defined as the sum of the differences in amino-acid properties caused by single-base changes from each codon to each other codon. However, the extent to which organisms optimize the genetic messages written in this code has been far less studied. We tested whether codon and amino-acid usages from 457 bacteria, 264 eukaryotes, and 33 archaea minimize errors compared to random usages, and whether changes in genome G+C content influence these error values.

Results

We tested the hypotheses that organisms choose their codon usage to minimize errors, and that the large observed variation in G+C content in coding sequences, but the low variation in G+U or G+A content, is due to differences in the effects of variation along these axes on the error value. Surprisingly, the biological distribution of error values has far lower variance than randomized error values, but error values of actual codon and amino-acid usages are actually greater than would be expected by chance.

Conclusion

These unexpected findings suggest that selection against translation error has not produced codon or amino-acid usages that minimize the effects of errors, and that even messages with very different nucleotide compositions somehow maintain a relatively constant error value. They raise the question: why do all known organisms use highly error-minimizing genetic codes, but fail to minimize the errors in the mRNA messages they encode?

     

False occurrences of functional motifs in protein sequences highlight evolutionary constraints

Background

False occurrences of functional motifs in protein sequences can be considered as random events due solely to the sequence composition of a proteome. Here we use a numerical approach to investigate the random appearance of functional motifs with the aim of addressing biological questions such as: How are organisms protected from undesirable occurrences of motifs otherwise selected for their functionality? Has the random appearance of functional motifs in protein sequences been affected during evolution?

Results

Here we analyse the occurrence of functional motifs in random sequences and compare it to that observed in biological proteomes; the behaviour of random motifs is also studied. Most motifs exhibit a number of false positives significantly similar to the number of times they appear in randomized proteomes (=expected number of false positives). Interestingly, about 3% of the analysed motifs show a different kind of behaviour and appear in biological proteomes less than they do in random sequences. In some of these cases, a mechanism of evolutionary negative selection is apparent; this helps to prevent unwanted functionalities which could interfere with cellular mechanisms.

Conclusion

Our thorough statistical and biological analysis showed that there are several mechanisms and evolutionary constraints both of which affect the appearance of functional motifs in protein sequences.

     

An EST resource for tilapia based on 17 normalized libraries and assembly of 116,899 sequence tags

Background

Large collections of expressed sequence tags (ESTs) are a fundamental resource for analysis of gene expression and annotation of genome sequences. We generated 116,899 ESTs from 17 normalized and two non-normalized cDNA libraries representing 16 tissues from tilapia, a cichlid fish widely used in aquaculture and biological research.

Results

The ESTs were assembled into 20,190 contigs and 36,028 singletons for a total of 56,218 unique sequences and a total assembled length of 35,168,415 bp. Over the whole project, a unique sequence was discovered for every 2.079 sequence reads. 17,722 (31.5%) of these unique sequences had significant BLAST hits (e-value < 10^-10) to the UniProt database.

Conclusion

Normalization of the cDNA pools with double-stranded nuclease allowed us to efficiently sequence a large collection of ESTs. These sequences are an important resource for studies of gene expression, comparative mapping and annotation of the forthcoming tilapia genome sequence.