BLT Stages

Abstract Bachelor Project FBT 2020-2021: Identification of Gardnerella strains from the culture collection of the Laboratory of Bacteriological Research

The aim of this research is to identify the Gardnerella species in the LBR collection. There are thirteen genomospecies, of which four have already been described, namely G. vaginalis, G. leopoldii, G. piotii and G. swidsinskii. Identification was done using different methods. First, the genus Gardnerella was screened. Thereafter, the G. vaginalis strains were filtered out via qPCR and only non-vaginalis continued to be worked on. qPCR for the species already described had been performed on all strains. PCR was done to amplify, and this was visualized via gel electrophoresis. All strains of which a DNA fragment could be seen were sent for sequencing. In addition, a reference database was compiled via a previous study by LBR. The sequences were compared to the reference database. A phylogenetic tree was set up. This tree was compared with the results of qPCR. From this it could be concluded that strains of G. piotii, G. swidsinskii, G. leopoldii and genomospecies 3, genomospecies 11, genomospecies 12 may be present in the LBR collection. No strains from genomospecies 7, genomospecies 8, genomospecies 9 and genomospecies 10 were found.

 

One of the most threatening concerns for hospitals is the outbreak of multidrug resistant organisms. Colonized patients need to be kept in isolation to prevent the spreading of the bacteria and are at risk of developing infections such as bacteremia. Outbreak analysis is essential for risk assessment, outbreak management and forecasting in the hospital. Using Whole Genome Sequencing (WGS) the bacterial isolates can be characterized and the sequencing data (fastqs) can be used to determine the relation between different patient or environmental samples. This is done with core genome or whole genome multilocus sequence typing (cgMLST or wgMLST). Commercially available platforms to do the analysis come at great expenses and are often a black box. By generating an own pipeline the UZ Gent hospital can lower the expenses and acquire relevant knowhow. To perform cgMLST analysis and minimum spanning tree (MST) generation the analysis pipeline (“OUTB8-analysis v1.0”) was created in Nextflow (v20.01.0.5264), see figure 1. OUTB8-analysis accepts paired and single end fastq files and requires a chewBBACA compatible cgMLST scheme as input. The pipeline starts with a quality control step of the raw data (FastQC v0.11.9, MultiQC v1.8) and continues with adaptor and quality trimming (fastp v0.20.0) and subsequent quality analysis. A de novo assembly is generated (megahit v1.2.9) and checked for quality (MetaQUAST v5.0.2). The simple MLST-type is determined using the PubMLST database (mlst v2.19.0). Extra assemblies can be added to the analysis with the “assem” option. Optionally (option “x”) the pipeline is ran with the given assemblies only without executing previous prepping steps. Actual cgMLST analysis is performed with the chewBBACA software (v2.5.4). The resulting table (tsv-file) is cleaned up and send to an R script for the generation of a minimum spanning tree. This tree enables quick visual detection of outbreaks within the analyzed samples. cgMLST schemes are not freely available for all bacteria, accordingly the scheme pipeline “OUTB8-scheme v1.0” was developed. This pipeline created in Nextflow (v20.01.0.5264) generates a prodigal training file and an chewBBACA compatible wgMLST and cgMLST scheme based on a given txid. Because a set of reference assemblies is needed to perform cg/wgMLST scheme creation, the pipeline searches for an existing accession file in the local MLST-database or generates a new file from NCBI using esearch (v13.3). The number of assemblies used to generate the scheme can be limited with the option “count” if desired. Selected fasta files are downloaded with bit-dl-ncbi-assemblies. A prodigal (v2.6.3) training file is created from a generated multifasta. Next the wgMLST scheme is created using the reference assemblies with chewBBACA software (v2.5.4). After allele calling, paralogs are removed and quality is checked. The cgMLST scheme can be generated with all loci present in a chosen percentage of the reference assemblies (option “perc”, default 95%). The same pipeline can also be used to determine the txid of several fastq files. This is done with kraken2 (v2.0.8-beta) on mini-fastq files of 10,000 reads to reduce computational power. Both pipelines were tested on freely available data of Klebsiella pneumonia outbreaks. The pipelines have shown to be adaptable and easy to use. In short these two pipelines provide an easy an adaptable way of analyzing outbreak samples starting from raw fastq files. In the future the OUTB8-analysis pipeline will be improved by choice of assembler and addition of a scaffolder. Also the OUTB8-scheme pipeline will include the adaption of already existing schemes to chewBBACA compatible schemes. After further validation with locally available data of several bacteria species and version control the pipeline will be used in UZGent for bacterial outbreak management.

Abstract 2 advanced bachelor of bioinformatics 2019-2020: Comparative genome analysis of Gardnerella species with a focus on taxonomy

Bacterial vaginosis is a disturbance of the healthy vaginal microbiome where the lactobacilli are replaced with anaerobes such as Gardnerella vaginalis.  This increases the acidity in the vagina. Under normal circumstances, the lactic acid-producing lactobacilli ensure a fairly high acidity in the natural environment of the vagina. Bacterial vaginosis is associated with premature birth and increased incidence of sexually transmitted infections including HIV.

Gardnerella vaginalis, the key pathogen in BV, has recently been shown to actually comprise 13 different species, based on the comparison of 81 genomes. Of these 13 species, some were suggested to be more virulant than others. My task in at the traineeship was to update this taxonomy. For this all the genomes of Gardnerella vaginalis from NCBI were downloaded. From this information different kinds of data were extracted with the use of python to in the first place control if the sequences were indeed of Gardnerella vaginalis. This was done by using the 16S ribosomal RNA region of the genome and if this was not present in the data other genes were used for this quality control. Control gene was used in a blast, if the top hit was Gardnerella vaginalis the genome passed the quality control else the genome could not be classified as Gardnerella vaginalis. The second part of the traineeship was the update of the figure where the ANI (average nucleotide identity) and the DDH (DNA-DNA-hybridization) are placed next to each other for proper subdivision of the different subspecies of Gardnerella vaginalis. 

Both the average nucleotide identity and DNA-DNA-hybridization used the genomes provided from NCBI. The results were gathered and placed in excel and filtered. 

The process of gathering the 16S ribosomal RNA, getting the genome files and using the ANI tool were atomized using python. Here the only input needed is the csv file from NCBI that can be downloaded when going to the genome page of a bacteria. This script is thus also usable for researching other bacteria than Gardnerella vaginalis. The DDH and the quality control and the filtering of the results to get the wanted structure are done manually due to a lack of time but will be atomized by the next trainee.

When looking at this figure different groups are visible which separate the different genomes in subspecies. This can be important for the next step when looking at the virulent genes, the virulent genes can differ in these groups and thus the activity of the specific made bacteriophage can differ between these groups.

The last phase of the internship was to use the virulent genes found in publications and provided from the internship to determine which genes are present on the genomes of these bacteria and which genes to use in the phage therapy to target these bacteria without interacting with the other bacteria present in the microbiome. In this way specific bacteria species can be eradicated from a microbiome to restore balance.

 

Abstract Bachelor Project FBT 2018-2019: Development and optimization of ddPCR for detection and qualification of a SNP linked to anthelmintic resistance in Trichuris trichiura

Soil-transmitted helminths (Ascaris lumbricoides, Trichuris trichiura and the hookworms Ancylostoma duodenale and Necator americanus) are a group of intestinal parasitic nematodes that infect one fifth of the world’s population and cause significant morbidity. Preventive chemotherapy with benzimidazoles (BZ, i.e. albendazole or mebendazole) is implemented as the main strategy for controlling Soil-transmitted helminthiasis. However, there is grounded fear that drug resistance may develop as a result of (1) the high degree of drug pressure, (2) the suboptimal doses and (3) the reliance on only two drugs that have a common mode of action.

Experience from the veterinary field has further ratified these concerns. In veterinary nematodes, BZ-resistance is caused by single nucleotide polymorphisms (SNPs) located in the β-tubulin isotype 1 gene at codon position 167, 198 and/or 200. Therefore, the development of a system to monitor the emergence and spread of anthelmintic resistance is crucial. In this study, a novel competitive digital droplet PCR (ddPCR) assay was designed for the detection of both the wildtype and the SNP 200 in the β-tubulin isotype 1 gene of T. trichiura. This species was chosen based on data that indicate that BZ efficacy against T. trichiura is unsatisfactory, hence likely the biggest chance of finding resistance. The newly developed ddPCR assays were first optimized using quantitative Polymerase Chain Reaction (qPCR). The results indicate that the final assay allows specific detection and quantification of mutant and wild-type targets.

Further optimization is necessary to make the assay even more efficient, as it may be valuable for the detection and monitoring of anthelminthic resistance in the future.