Dataset Information

Improving de novo sequence assembly using machine learning and comparative genomics for overlap correction.

ABSTRACT: BACKGROUND: With the rapid expansion of DNA sequencing databases, it is now feasible to identify relevant information from prior sequencing projects and completed genomes and apply it to de novo sequencing of new organisms. As an example, this paper demonstrates how such extra information can be used to improve de novo assemblies by augmenting the overlapping step. Finding all pairs of overlapping reads is a key task in many genome assemblers, and to this end, highly efficient algorithms have been developed to find alignments in large collections of sequences. It is well known that due to repeated sequences, many aligned pairs of reads nevertheless do not overlap. But no overlapping algorithm to date takes a rigorous approach to separating aligned but non-overlapping read pairs from true overlaps. RESULTS: We present an approach that extends the Minimus assembler by a data driven step to classify overlaps as true or false prior to contig construction. We trained several different classification models within the Weka framework using various statistics derived from overlaps of reads available from prior sequencing projects. These statistics included percent mismatch and k-mer frequencies within the overlaps as well as a comparative genomics score derived from mapping reads to multiple reference genomes. We show that in real whole-genome sequencing data from the E. coli and S. aureus genomes, by providing a curated set of overlaps to the contigging phase of the assembler, we nearly doubled the median contig length (N50) without sacrificing coverage of the genome or increasing the number of mis-assemblies. CONCLUSIONS: Machine learning methods that use comparative and non-comparative features to classify overlaps as true or false can be used to improve the quality of a sequence assembly.

SUBMITTER: Palmer LE

PROVIDER: S-EPMC2824677 | biostudies-literature | 2010

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Improving de novo sequence assembly using machine learning and comparative genomics for overlap correction.

Palmer Lance E LE Dejori Mathaeus M Bolanos Randall R Fasulo Daniel D

BMC bioinformatics 20100115

<h4>Background</h4>With the rapid expansion of DNA sequencing databases, it is now feasible to identify relevant information from prior sequencing projects and completed genomes and apply it to de novo sequencing of new organisms. As an example, this paper demonstrates how such extra information can be used to improve de novo assemblies by augmenting the overlapping step. Finding all pairs of overlapping reads is a key task in many genome assemblers, and to this end, highly efficient algorithms ...[more]

PMID: 20078885

Similar Datasets

Project description:BACKGROUND: Common carp (Cyprinus carpio) is one of the most important aquaculture species of Cyprinidae with an annual global production of 3.4 million tons, accounting for nearly 14% of the freshwater aquaculture production in the world. Due to the economical and ecological importance of common carp, genomic data are eagerly needed for genetic improvement purpose. However, there is still no sufficient transcriptome data available. The objective of the project is to sequence transcriptome deeply and provide well-assembled transcriptome sequences to common carp research community. RESULT: Transcriptome sequencing of common carp was performed using Roche 454 platform. A total of 1,418,591 clean ESTs were collected and assembled into 36,811 cDNA contigs, with average length of 888 bp and N50 length of 1,002 bp. Annotation was performed and a total of 19,165 unique proteins were identified from assembled contigs. Gene ontology and KEGG analysis were performed and classified all contigs into functional categories for understanding gene functions and regulation pathways. Open Reading Frames (ORFs) were detected from 29,869 (81.1%) contigs with an average ORF length of 763 bp. From these contigs, 9,625 full-length cDNAs were identified with sequence length from 201 bp to 9,956 bp. Comparative analysis revealed that 27,693(75.2%) contigs have significant similarity to zebrafish Refseq proteins, and 24,371(66.2%), 24,501(66.5%) and 25,025(70.0%) to teraodon, medaka and three-spined stickleback refseq proteins. A total of 2,064 microsatellites were initially identified from 1,730 contigs, and 1,639 unique sequences had sufficient flanking sequences on both sides for primer design. CONCLUSION: The transcriptome of common carp had been deep sequenced, de novo assembled and characterized, providing the valuable resource for better understanding of common carp genome. The transcriptome data will facilitate future functional studies on common carp genome, and gradually apply in breeding programs of common carp, as well as closely related other Cyprinids.

Project description:BackgroundSeveral standalone error correction tools have been proposed to correct sequencing errors in Illumina data in order to facilitate de novo genome assembly. However, in a recent survey, we showed that state-of-the-art assemblers often did not benefit from this pre-correction step. We found that many error correction tools introduce new errors in reads that overlap highly repetitive DNA regions such as low-complexity patterns or short homopolymers, ultimately leading to a more fragmented assembly.ResultsWe propose BrownieCorrector, an error correction tool for Illumina sequencing data that focuses on the correction of only those reads that overlap short DNA patterns that are highly repetitive in the genome. BrownieCorrector extracts all reads that contain such a pattern and clusters them into different groups using a community detection algorithm that takes into account both the sequence similarity between overlapping reads and their respective paired-end reads. Each cluster holds reads that originate from the same genomic region and hence each cluster can be corrected individually, thus providing a consistent correction for all reads within that cluster.ConclusionsBrownieCorrector is benchmarked using six real Illumina datasets for different eukaryotic genomes. The prior use of BrownieCorrector improves assembly results over the use of uncorrected reads in all cases. In comparison with other error correction tools, BrownieCorrector leads to the best assembly results in most cases even though less than 2% of the reads within a dataset are corrected. Additionally, we investigate the impact of error correction on hybrid assembly where the corrected Illumina reads are supplemented with PacBio data. Our results confirm that BrownieCorrector improves the quality of hybrid genome assembly as well. BrownieCorrector is written in standard C++11 and released under GPL license. BrownieCorrector relies on multithreading to take advantage of multi-core/multi-CPU systems. The source code is available at https://github.com/biointec/browniecorrector .

Dataset Information

Improving de novo sequence assembly using machine learning and comparative genomics for overlap correction.

Publications

Improving de novo sequence assembly using machine learning and comparative genomics for overlap correction.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets