Large scale genotype-phenotype correlation analysis based on phylogenetic trees

Machine learning and phylogenetic analysis allow for predicting antibiotic resistance in M. tuberculosis

BACKGROUND

Antimicrobial resistance (AMR) poses a significant global health threat, and an accurate prediction of bacterial resistance patterns is critical for effective treatment and control strategies. In recent years, machine learning (ML) approaches have emerged as powerful tools for analyzing large-scale bacterial AMR data. However, ML methods often ignore evolutionary relationships among bacterial strains, which can greatly impact performance of the ML methods, especially if resistance-associated features are attempted to be detected. Genome-wide association studies (GWAS) methods like linear mixed models accounts for the evolutionary relationships in bacteria, but they uncover only highly significant variants which have already been reported in literature.

RESULTS

In this work, we introduce a novel phylogeny-related parallelism score (PRPS), which measures whether a certain feature is correlated with the population structure of a set of samples. We demonstrate that PRPS can be used, in combination with SVM- and random forest-based models, to reduce the number of features in the analysis, while simultaneously increasing models' performance. We applied our pipeline to publicly available AMR data from PATRIC database for Mycobacterium tuberculosis against six common antibiotics.

CONCLUSIONS

Using our pipeline, we re-discovered known resistance-associated mutations as well as new candidate mutations which can be related to resistance and not previously reported in the literature. We demonstrated that taking into account phylogenetic relationships not only improves the model performance, but also yields more biologically relevant predicted most contributing resistance markers.

Forest and Trees: Exploring Bacterial Virulence with Genome-wide Association Studies and Machine Learning

The advent of inexpensive and rapid sequencing technologies has allowed bacterial whole-genome sequences to be generated at an unprecedented pace. This wealth of information has revealed an unanticipated degree of strain-to-strain genetic diversity within many bacterial species. Awareness of this genetic heterogeneity has corresponded with a greater appreciation of intraspecies variation in virulence. A number of comparative genomic strategies have been developed to link these genotypic and pathogenic differences with the aim of discovering novel virulence factors. Here, we review recent advances in comparative genomic approaches to identify bacterial virulence determinants, with a focus on genome-wide association studies and machine learning.

Current Affairs of Microbial Genome-Wide Association Studies: Approaches, Bottlenecks and Analytical Pitfalls

Microbial genome-wide association studies (mGWAS) are a new and exciting research field that is adapting human GWAS methods to understand how variations in microbial genomes affect host or pathogen phenotypes, such as drug resistance, virulence, host specificity and prognosis. Several computational tools and methods have been developed or adapted from human GWAS to facilitate the discovery of novel mutations and structural variations that are associated with the phenotypes of interest. However, no comprehensive, end-to-end, user-friendly tool is currently available. The development of a broadly applicable pipeline presents a real opportunity among computational biologists. Here, (i) we review the prominent and promising tools, (ii) discuss analytical pitfalls and bottlenecks in mGWAS, (iii) provide insights into the selection of appropriate tools, (iv) highlight the gaps that still need to be filled and how users and developers can work together to overcome these bottlenecks. Use of mGWAS research can inform drug repositioning decisions as well as accelerate the discovery and development of more effective vaccines and antimicrobials for pressing infectious diseases of global health significance, such as HIV, TB, influenza, and malaria.

Phylogenetic Concepts and Tools Applied to Epidemiologic Investigations of Infectious Diseases

In this review, which is a part of the Microbiology Spectrum Curated Collection: Advances in Molecular Epidemiology of Infectious Diseases, I present an overview of the principles used to classify organisms in the field of phylogenetics, highlight the methods used to infer the interrelationships of organisms, and summarize how these concepts are applied to molecular epidemiologic analyses. I present steps in analyses that come downstream of the assembly of a set of genomes or genes and the production of a multiple-sequence alignment or other matrices of putative orthologs for comparison. I focus on the history of the problem of phylogenetic reconstruction and debates within the field about the most appropriate methods. I illustrate methods that bridge the gap between molecular epidemiology and traditional epidemiology, including phylogenetic character evolution and geographic visualization. Finally, I provide practical advice on how to conduct an example analysis in the appendix. *This article is part of a curated collection.

Cladograms with Path to Event (ClaPTE): a novel algorithm to detect associations between genotypes or phenotypes using phylogenies

BACKGROUND

Associations between genotype and phenotype provide insight into the evolution of pathogenesis, drug resistance, and the spread of pathogens between hosts. However, common ancestry can lead to apparent associations between biologically unrelated features. The novel method Cladograms with Path to Event (ClaPTE) detects associations between character-pairs (either a pair of mutations or a mutation paired with a phenotype) while adjusting for common ancestry, using phylogenetic trees.

METHODS

ClaPTE tests for character-pairs changing close together on the phylogenetic tree, consistent with an associated character-pair. ClaPTE is compared to three existing methods (independent contrasts, mixed model, and likelihood ratio) to detect character-pair associations adjusted for common ancestry. Comparisons utilize simulations on gene trees for: HIV Env, HIV promoter, and bacterial DnaJ and GuaB; and case studies for Oseltamavir resistance in Influenza, and for DnaJ and GuaB. Simulated data include both true-positive/associated character-pairs, and true-negative/not-associated character-pairs, used to assess type I (frequency of p-values in true-negatives) and type II (sensitivity to true-positives) error control.

RESULTS AND CONCLUSIONS

ClaPTE has competitive sensitivity and better type I error control than existing methods. In the Influenza/Oseltamavir case study, ClaPTE reports no new permissive mutations but detects associations between adjacent (in primary sequence) amino acid positions which other methods miss. In the DnaJ and GuaB case study, ClaPTE reports more frequent associations between positions both from the same protein family than between positions from different families, in contrast to other methods. In both case studies, the results from ClaPTE are biologically plausible.

GWAMAR: genome-wide assessment of mutations associated with drug resistance in bacteria

Background

Development of drug resistance in bacteria causes antibiotic therapies to be less effective and more costly. Moreover, our understanding of the process remains incomplete. One promising approach to improve our understanding of how resistance is being acquired is to use whole-genome comparative approaches for detection of drug resistance-associated mutations.

Results

We present GWAMAR, a tool we have developed for detecting of drug resistance-associated mutations in bacteria through comparative analysis of whole-genome sequences. The pipeline of GWAMAR comprises several steps. First, for a set of closely related bacterial genomes, it employs eCAMBer to identify homologous gene families. Second, based on multiple alignments of the gene families, it identifies mutations among the strains of interest. Third, it calculates several statistics to identify which mutations are the most associated with drug resistance.

Conclusions

Based on our analysis of two large datasets retrieved from publicly available data for M. tuberculosis, we identified a set of novel putative drug resistance-associated mutations. As a part of this work, we present also an application of our tool to detect putative compensatory mutations.

Explaining microbial phenotypes on a genomic scale: GWAS for microbes

There is an increasing availability of complete or draft genome sequences for microbial organisms. These data form a potentially valuable resource for genotype-phenotype association and gene function prediction, provided that phenotypes are consistently annotated for all the sequenced strains. In this review, we address the requirements for successful gene-trait matching. We outline a basic protocol for microbial functional genomics, including genome assembly, annotation of genotypes (including single nucleotide polymorphisms, orthologous groups and prophages), data pre-processing, genotype-phenotype association, visualization and interpretation of results. The methodologies for association described herein can be applied to other data types, opening up possibilities to analyze transcriptome-phenotype associations, and correlate microbial population structure or activity, as measured by metagenomics, to environmental parameters.

An approach to identifying drug resistance associated mutations in bacterial strains

Background

Drug resistance in bacterial pathogens is an increasing problem, which stimulates research. However, our understanding of drug resistance mechanisms remains incomplete. Fortunately, the fast-growing number of fully sequenced bacterial strains now enables us to develop new methods to identify mutations associated with drug resistance.

Results

We present a new comparative approach to identify genes and mutations that are likely to be associated with drug resistance mechanisms. In order to test the approach, we collected genotype and phenotype data of 100 fully sequenced strains of S. aureus and 10 commonly used drugs. Then, applying the method, we re-discovered the most common genetic determinants of drug resistance and identified some novel putative associations.

Conclusions

Firstly, the collected data may help other researchers to develop and verify similar techniques. Secondly, the proposed method is successful in identifying drug resistance determinants. Thirdly, the in-silico identified genetic mutations, which are putatively involved in drug resistance mechanisms, may increase our understanding of the drug resistance mechanisms.

PhenoLink--a web-tool for linking phenotype to ~omics data for bacteria: application to gene-trait matching for Lactobacillus plantarum strains

BACKGROUND

Linking phenotypes to high-throughput molecular biology information generated by ~omics technologies allows revealing cellular mechanisms underlying an organism's phenotype. ~Omics datasets are often very large and noisy with many features (e.g., genes, metabolite abundances). Thus, associating phenotypes to ~omics data requires an approach that is robust to noise and can handle large and diverse data sets.

RESULTS

We developed a web-tool PhenoLink (http://bamics2.cmbi.ru.nl/websoftware/phenolink/) that links phenotype to ~omics data sets using well-established as well new techniques. PhenoLink imputes missing values and preprocesses input data (i) to decrease inherent noise in the data and (ii) to counterbalance pitfalls of the Random Forest algorithm, on which feature (e.g., gene) selection is based. Preprocessed data is used in feature (e.g., gene) selection to identify relations to phenotypes. We applied PhenoLink to identify gene-phenotype relations based on the presence/absence of 2847 genes in 42 Lactobacillus plantarum strains and phenotypic measurements of these strains in several experimental conditions, including growth on sugars and nitrogen-dioxide production. Genes were ranked based on their importance (predictive value) to correctly predict the phenotype of a given strain. In addition to known gene to phenotype relations we also found novel relations.

CONCLUSIONS

PhenoLink is an easily accessible web-tool to facilitate identifying relations from large and often noisy phenotype and ~omics datasets. Visualization of links to phenotypes offered in PhenoLink allows prioritizing links, finding relations between features, finding relations between phenotypes, and identifying outliers in phenotype data. PhenoLink can be used to uncover phenotype links to a multitude of ~omics data, e.g., gene presence/absence (determined by e.g.: CGH or next-generation sequencing), gene expression (determined by e.g.: microarrays or RNA-seq), or metabolite abundance (determined by e.g.: GC-MS).

Exclusion of Kif1c as a candidate gene for anthrax toxin susceptibility

Different strains of mice possess varying degrees of susceptibility to anthrax lethal toxin (LT). Previous studies have suggested a responsible locus Ltxs1 that contains 10 or more known genes, but functional relevance has been reported for two genes, Kif1c and Nalp1b. In this study, we attempted to determine the involvement of Kif1c in anthrax susceptibility using Kif1c knockout mice. We established Kif1c knockout mice with LT-sensitive 129/Sv-derived embryonic stem cells followed by 13 backcrosses with LT-resistant C57BL/6J mice (B6) to be congenic. These knockout mice and their primary macrophages showed significantly higher sensitivity to LT than wild-type B6. However, when we replaced the remaining 129/Sv genome adjacent to the targeted Kif1c locus with the B6 genome, this sensitivity was lost. This suggested that the sensitivity to LT in the originally established Kif1c knockout mice was not due to the loss of the Kif1c gene, but was because of the presence of the 129/Sv-derived genes adjacent to the disrupted Kif1c locus. Thus, Kif1c was excluded as a candidate anthrax susceptibility gene.

Evolution of drug resistance in multiple distinct lineages of H5N1 avian influenza

Some predict that influenza A H5N1 will be the cause of a pandemic among humans. In preparation for such an event, many governments and organizations have stockpiled antiviral drugs such as oseltamivir (Tamiflu). However, it is known that multiple lineages of H5N1 are already resistant to another class of drugs, adamantane derivatives, and a few lineages are resistant to oseltamivir. What is less well understood is the evolutionary history of the mutations that confer drug resistance in the H5N1 population. In order to address this gap, we conducted phylogenetic analyses of 676 genomic sequences of H5N1 and used the resulting hypotheses as a basis for asking 3 molecular evolutionary questions: (1) Have drug-resistant genotypes arisen in distinct lineages of H5N1 through point mutation or through reassortment? (2) Is there evidence for positive selection on the codons that lead to drug resistance? (3) Is there evidence for covariation between positions in the genome that confer resistance to drugs and other positions, unrelated to drug resistance, that may be under selection for other phenotypes? We also examine how drug-resistant lineages proliferate across the landscape by projecting or phylogenetic analysis onto a virtual globe. Our results for H5N1 show that in most cases drug resistance has arisen by independent point mutations rather than reassortment or covariation. Furthermore, we found that some codons that mediate resistance to adamantane derivatives are under positive selection, but did not find positive selection on codons that mediate resistance to oseltamivir. Together, our phylogenetic methods, molecular evolutionary analyses, and geographic visualization provide a framework for analysis of globally distributed genomic data that can be used to monitor the evolution of drug resistance.

US and Scottish health professionals' attitudes toward DNA biobanking

BACKGROUND

The authors define a DNA biobank as a repository of genetic information correlated with patient medical records. DNA biobanks may assist in the research and identification of genetic factors influencing disease and drug interactions, but may raise ethical issues. How healthcare providers perceive DNA biobanks is unknown.

OBJECTIVES

To determine how useful healthcare professionals believe DNA biobanks will be and whether these attitudes differ between private and socialized healthcare systems.

DESIGN

The authors surveyed 200 healthcare professionals, including research and non-research focused doctors, nurses and other staff from medical centers and independent practice in both the United States and Scotland. The survey included fifteen items evaluated for general receptiveness toward biobanks, presumed usefulness of biobanks and perceived attitudes in recruiting patients for a biobank.

MEASUREMENTS

A total of 81 (45%) of 179 eligible participants responded: 41 from the U.S. and 40 from Scotland. Of these respondents, most (70%) were from academic centers.

RESULTS

Results indicate that there is a broadly favorable attitude in both locations toward the creation of a DNA biobank (83%) and its perceived benefit (75%). This enthusiasm is tempered in Scotland when respondents evaluated their comfort in consenting patients for entry into a biobank; 16 of 40 respondents (40%) were uncomfortable doing so, representing a significant difference from those in the U.S. (p=0.001).

CONCLUSIONS

Despite systematic differences in healthcare practice between the U.S. and Scotland, health care professionals in both nations believe DNA biobanks will be useful in curing disease. This finding appears to support further development of such a research tool.