wgMLST as a standardized tool for assessing the quality of genome assembly data

Dmitriy Babenko, Ilya Azizov, Mark Alexander Toleman

2016

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.1016/j.ijid.2016.02.714
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Background: Continued advances in next-generation sequencing (NGS) technologies is accompanied with the development of many whole genome assembly approaches to convert the small sequences (reads) into larger regions (contigs/scaffolds). However, none of these is perfect. Up to now, genome assembly data is compared under standard statistics (N50, coverage, contig sizes, number bases etc.) and there is no commonly accepted and standardized method for comparison and assessing the assembly data Methods & Materials: The raw data for S. aureus SA957 (paired end sequencing - SRR497751) produced by Illumina platform have been download from European Nucleotide Archive. Software such as Tadpole, Velvet, CLC genomic workbench, SeqMan NGen for de novo assembly and Bowtie2, BWA and CLC genomic workbench for mapping to reference have been used under default options to produce contigs/consensus. The assessing of the quality of genome assembly have been performed with using wgMLST implemented in SeqSphere software (Ridom). Concordace of genome assembly data was estimated with Rand index. Results: Seven genomes have been assembled using de novo and reference mapping methods. The basic statistics of de novo assembled data showed that CLC genomic workbench tool gave the best assembly sequence. Statistic parameters of Tadpole assembly were less well in comparison with other (Table 1). wgMLST analysis based on 2787 genes was performed on seven assembled genomes. As a result, CLC, Velvet and SeqMan Ngen assembly allowed to determine more than 2500 genes while Tadpole assembly with average contig length 1378 bp could identify 1560 out 2787 genes (table 2). Reference mapping assembly reviealed hign concordance (98-99%) between results. Minimum spanning tree clustered reference mapping results (picture1).View Large Image Figure ViewerDownload Hi-res image Download (PPT) Conclusion: A standardized gene-by-gene wgMLST approach allows assessing not only the quality but also quantitative estimation of genome assembly data. Based on this approach the genomes assembled with different software can be compared and clustered to find approaches that give similar results. wgMLST allows to find the discrepancies on gene level as well.