Genome assemblies and respective wg/cgMLST profiles of a diverse dataset comprising 1,999 Escherichia coli isolates

Dataset

This dataset comprises the genome assemblies and respective 7,601-loci whole-genome (wg) Multiple Locus Sequence Type (MLST) profiles [INNUENDO schema (Llarena et al. 2018) available in chewie-NS (Mamede et al. 2022)] of a final set of 1,999 Escherichia coli samples selected among the Whole-Genome Sequencing (WGS) data publicly available in the European Nucleotide Archive (ENA) or in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) at the beginning of the analysis (November 2021). This set of samples was carefully selected to cover a wide genetic diversity (assessed in terms of serotype). In total, 411 different serotypes are represented in this dataset, with O157:H7 being the most represented one, corresponding to 37.1% of the dataset.

File “Ec_metadata.xlsx” contains metadata information for each isolate, including ENA/SRA accession number, BioProject and in-silico MLST ST and serotype.

The directory “assemblies/” contains all the genome assemblies (.fasta format) of each isolate presented in the metadata file. 

The file “profiles/Ec_profiles_wgMLST.tsv” corresponds to a tab separated file with the 7,601-loci wgMLST profiles of each isolate presented in the metadata file. The files “profiles/Ec_profiles_cgMLST_95.tsv”, “profiles/Ec_profiles_cgMLST_98.tsv” and “profiles/Ec_profiles_cgMLST_100.tsv” correspond to a 2,826-loci, 2,704-loci and 465-loci cgMLST profiles of each isolate presented in the metadata file, respectively. These profiles were determined as explained below.

Dataset selection and curation

With the objective of creating a diverse dataset of E. coli genome assemblies, we collected information about the genetic diversity (serotype) of the isolates available at Enterobase database in the beginning of this analysis (November 2021) and in other previous works. Based on this information, we selected an initial dataset comprising 2,688 samples associated with three BioProjects (PRJNA230969, PRJEB27020 and PRJNA248042). Their WGS data was downloaded from ENA/SRA with fastq-dl v1.0.6. Read quality control, trimming and assembly were performed with the Aquamis v1.3.9 (Deneke et al. 2021) using default parameters. Assembly quality control (QC), including contamination assessment, as well as MLST ST determination were performed with the same pipeline. All genome assemblies passing the QC were included in the final dataset. Among the others, we noticed that a considerable proportion of assemblies was flagged as “QC fail” exclusively due to the “NumContamSNVs” parameter, suggesting that this setting might have been too strict. After manual inspection of a random subset, assemblies for which the percentage of reads corresponding to the correct species was >98% were recovered and integrated in the final dataset (those samples are labeled in the Metadata file). In total, 1,999 isolates passed this curation step and were included in the final dataset. In-silico serotyping was performed with seq_typing v2.2. wgMLST profiles of each of these isolates were determined with chewBBACA v2.8.5 (Silva et al. 2018), using the 7,601-loci INNUENDO schema available in chewie-NS (Llarena et al. 2018; Mamede et al. 2022) and downloaded on May 31st, 2022. Three cgMLST schemas were obtained with ReporTree v1.0.0 (Mixão et al. 2022) using the 7,601-loci wgMLST profiles of the 1,999 isolates as input and setting distinct “--site-inclusion” thresholds: 0.95, 0.98 and 1.0 (i.e., keep schema loci called in at least 95%, 98% and 100% of the samples, resulting in a 2,826-loci, 2,704-loci and 465-loci allelic matrices, respectively).

Acknowledgements

We thank the National Distributed Computing Infrastructure of Portugal (INCD) for providing the necessary resources to run the genome assemblies. INCD was funded by FCT and FEDER under the project 22153-01/SAICT/2016.