bioRxiv ScienceSearch

Biology subjects

Essex, M.

Publications and source records attributed to Essex, M..

2 recordsLinked to original sources

SIAMCAT: user-friendly and versatile machine learning workflows for statistically rigorous microbiome analyses

The human microbiome is increasingly mined for diagnostic and therapeutic biomarkers using machine learning (ML). However, metagenomics-specific software is scarce and overoptimistic evaluation and limited cross-study generalization are prevailing issues. To address these, we developed SIAMCAT, a versatile R toolbox for ML-based comparative metagenomics. We demonstrate its capabilities in a meta-analysis of fecal metagenomic studies (10,803 samples). When naively transferred across studies, ML models lost accuracy and disease specificity, which could however be resolved by a novel training set augmentation strategy. This revealed some biomarkers to be disease-specific, others shared across multiple conditions. SIAMCAT is freely available from siamcat.embl.de.

bioinformatics

PANDAA-monium: Intentional violations of conventional qPCR design enables rapid, HIV-1 subtype-independent drug resistance SNP detection

Global efforts to ensure that 90% of all HIV-infected people receiving antiretroviral therapy (ART) will be virally suppressed by 2020 could be crippled by increases in acquired and transmitted HIV drug resistance (HIVDR), which challenge ART efficacy. The long-term sustainability of ART treatment programs is contingent on effective HIVDR monitoring yet current Sanger sequencing genotypic resistance tests are inadequate for large-scale implementation in low- and middle-income countries (LMICs). A simple, rapid, affordable HIVDR diagnostic would radically improve the treatment paradigm in LMICs by facilitating informed clinical decision-making upon ART failure. Although point mutation assays can be broadly deployed in this context, the primary challenge arises from extensive sequence variation surrounding targeted drug resistance mutations (DRMs). Here, we systematically and intentionally violate the canonical principles of qPCR design to develop a novel assay, Pan-Degenerate Amplification and Adaptation (PANDAA), that mitigates the impact of DRM-proximal secondary polymorphisms on probe-based qPCR performance to enable subtype-independent, focused resistance genotyping. Using extremely degenerate primers with 3 termini overlapping the probe-binding site, the HIV-1 genome is adapted through site-directed mutagenesis to replace secondary polymorphisms flanking the target DRM during the initial qPCR cycles. We show that PANDAA can quantify key HIV DRMs present at [≥]5% and has diagnostic sensitivity and specificity of 96.9% and 97.5%, respectively, to detect DRMs associated with ART failure. PANDAA is an innovative solution for HIVDR genotyping and is an advancement in qPCR technology that could be applicable to any scenario where target-proximal genetic variability has been a roadblock in diagnostic development.

molecular biology