About
Genome-Wide Association Studies on Antifungal Resistance in Aspergillus fumigatus
YEAR: 2021
SKILLS: GWAS | SNP Analysis | Statistical Genetics | Bioinformatics | R | Python
PURPOSE: Pipeline initially developed for course project at McMaster University; subsequently extended and applied at full scale during MSc thesis work in Dr. Jianping Xu's lab. Becoming the methodological foundation for 3 first authorship papers and a conference presentation.
LINKS: Github
Background:
Aspergillus fumigatus is a ubiquitous saprophytic mold and the leading cause of invasive aspergillosis, a life-threatening infection in immunocompromised patients with mortality rates ranging from 30-95%. Triazoles (including itraconazole and voriconazole) are used as first-line therapy, but rising worldwide triazole resistance has elevated the importance of second-line agents like amphotericin B (AMB). Amphotericin B is a polyene antifungal agent that has serves as a critical dalvage therapy for refractory invasive aspergillosis and is recommended as primary therapy in regions with ≥10% environmental triazole resistance. Patients with invasive aspergillosis caused by triazole-resistant isolates carry an ~88% mortality rate, making AMB efficacy critical.
AMB resistance has historically been uncommon for a fungicidal agent, but recent surveillance has reported strikingly high resistance rates (MIC ≥ 2 mg/L) in specific geographic populations - like 96.4% in Hamilton, Canada and 27% in Campinas, Brazil - and the reasons for this remain unclear. AMB's molecular mechanisms in A. fumigatus are also poorly characterized and the genetic determinants of resistance remain largely unexplored beyond a handful of candidate genes.
This project conducts an genome-wide association study to investigate AMB-resistance-associated loci, using clinical and environmental A. fumigatus isolates with measured AMB MIC values.
Methods & Approach:
End-to-end GWAS pipeline demonstrated on 32 A. fumigatus strains (the full analyses uses larger cohorts described in the linked publications)
Paired-end WGS reads from NCBI were QC'd, trimmed, and aligned to the Af293 reference genome with BWA-MEM, then joint-called per chromosome under a haploid model with FreeBayes. The resulting variants were filtered(vcftools + PLINK LD pruning), clonal strain groups were removed via IBD analysis, and AMB MIC phenotype data was normalized. Ran association testing under two complementary methods, PLINK linear regression and TASSEL GLM, using the top 10 principal components as covariates to control for population stratification. Results were visualized with Manhattan and Q–Q plots, and the top 10 SNPs (based on p-value) from each method were annotated and examined further with gene-function lookup.
Key Output:
This pipeline was first built as a course assignment for BIO722 (Introduction to Bioinformatic Methods) at McMaster University, where it was developed on a 32-strain subset to demonstrate the workflow end-to-end.
This pipeline was subsequently extended and applied at full scale during my MSc thesis work in Dr. Jianping Xu's lab, across both AMB and triazole resistance, and across larger panels of clinical and environmental isolates. This pipeline was the methodological foundation for 3 first authorship publications and a conference presentation listed below.
Peer-Reviewed Publications:
Conference Presentation:
"Genomic and Genetic Analyses of Antifungal Drug Resistance in Aspergillus fumigatus". 31st Fungal Genetics Conference, 2022 (program book). Poster presented by Dr. Jianping Xu; contents drafted by Yuying Fan based on MSc thesis work.