Evan B. Qu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Evan B. Qu has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Tami D. Lieberman
Institute for Medical Engineering and Sciences, Massachusetts Institute of Technology; Cambridge, MA 02139, USA
7 shared publications - Jacob S. Baker
Institute for Medical Engineering and Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
5 shared publications - Christopher P. Mancuso
Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, MA 02142, USA
4 shared publications - A. Delphine Tripp
Institute for Medical Engineering and Sciences, Massachusetts Institute of Technology; Cambridge
3 shared publications - Arolyn Conwill
Institute for Medical Engineering and Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
3 shared publications - Alyssa H. Mitchell
Institute for Medical Engineering and Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
2 shared publications
Latest Publications
- 6 July 2026
- Published by Cold Spring Harbor Laboratory in Genome Research
- Vol. 36 (9) , 1889-1901
High-accuracy SNV calling for bacterial isolates using deep learning with AccuSNV
Accurate detection of mutations within bacterial species is critical for fundamental studies of microbial evolution, reconstruction of transmission events, and identification of antimicrobial resistance mutations. Although many tools have been developed to identify single-nucleotide variants (SNVs) from whole-genome sequencing, they often suffer from high false-positive rates owing to the complexity of bacterial genomes and the need for different filtering cutoffs across sample types and sequencing depths. As data sets increase in size, the manual filtering required for high accuracy presents a significant obstacle. Here, we present AccuSNV, a novel deep learning–based tool for high-precision and automated bacterial SNV calling. Unlike traditional methods that process one sample at a time, AccuSNV leverages a convolutional neural network (CNN) that integrates alignment information across multiple samples, enhancing precision through learned across-sample patterns. We evaluate AccuSNV against seven popular SNV-calling tools using simulated data from six bacterial species with varied sequencing depths, numbers of isolates, mutations, and divergence levels. To further validate its real-world utility, we test AccuSNV on multiple curated bacterial data sets containing reported SNVs. In both simulated and real-world scenarios, AccuSNV consistently achieves the best performance. Moreover, AccuSNV provides comprehensive user-friendly downstream analysis modules and outputs, including mutation annotation information, phylogenetic inference, d N / d S calculations, and optional manual filtering. Together with the automated deep learning–based calling, these features make AccuSNV broadly accessible to users with different levels of computational expertise.
- 6 July 2026
- Published by Cold Spring Harbor Laboratory in Genome Research
- Vol. 36 (9) , 1889-1901
High-accuracy SNV calling for bacterial isolates using deep learning with AccuSNV
Accurate detection of mutations within bacterial species is critical for fundamental studies of microbial evolution, reconstruction of transmission events, and identification of antimicrobial resistance mutations. Although many tools have been developed to identify single-nucleotide variants (SNVs) from whole-genome sequencing, they often suffer from high false-positive rates owing to the complexity of bacterial genomes and the need for different filtering cutoffs across sample types and sequencing depths. As data sets increase in size, the manual filtering required for high accuracy presents a significant obstacle. Here, we present AccuSNV, a novel deep learning–based tool for high-precision and automated bacterial SNV calling. Unlike traditional methods that process one sample at a time, AccuSNV leverages a convolutional neural network (CNN) that integrates alignment information across multiple samples, enhancing precision through learned across-sample patterns. We evaluate AccuSNV against seven popular SNV-calling tools using simulated data from six bacterial species with varied sequencing depths, numbers of isolates, mutations, and divergence levels. To further validate its real-world utility, we test AccuSNV on multiple curated bacterial data sets containing reported SNVs. In both simulated and real-world scenarios, AccuSNV consistently achieves the best performance. Moreover, AccuSNV provides comprehensive user-friendly downstream analysis modules and outputs, including mutation annotation information, phylogenetic inference, d N / d S calculations, and optional manual filtering. Together with the automated deep learning–based calling, these features make AccuSNV broadly accessible to users with different levels of computational expertise.
- Open Access
High-accuracy SNV calling for bacterial isolates using deep learning with AccuSNV
Accurate detection of mutations within bacterial species is critical for fundamental studies of microbial evolution, reconstructing transmission events, and identifying antimicrobial resistance mutations. While many tools have been developed to identify single nucleotide variants (SNVs) from whole-genome sequencing, they often suffer from high false positive rates due to the complexity of bacterial genomes and the need for different filtering cutoffs across sample types and sequencing depths. As datasets increase in size, the manual filtering required for high accuracy presents a significant obstacle. Here, we present AccuSNV, a novel deep learning-based tool for high-precision and automated bacterial SNV calling. Unlike traditional methods that process one sample at a time, AccuSNV leverages a convolutional neural network (CNN) that integrates alignment information across multiple samples, enhancing precision through learned across-sample patterns. We evaluated AccuSNV against seven popular SNV calling tools using simulated data from six bacterial species with varied sequencing depths, numbers of isolates, mutations, and divergence levels. To further validate its real-world utility, we tested AccuSNV on multiple curated bacterial datasets containing reported SNVs. In both simulated and real-world scenarios, AccuSNV consistently achieved the best performance. Moreover, AccuSNV provides comprehensive user-friendly downstream analysis modules and outputs, including mutation annotation information, phylogenetic inference, dN/dS calculations, and optional manual filtering. Together with the automated deep learning–based calling, these features make AccuSNV broadly accessible to users with different levels of computational expertise.
Open AccessHigh-accuracy SNV calling for bacterial isolates using deep learning with AccuSNV
Accurate detection of mutations within bacterial species is critical for fundamental studies of microbial evolution, reconstructing transmission events, and identifying antimicrobial resistance mutations. While many tools have been developed to identify single nucleotide variants (SNVs) from whole-genome sequencing, they often suffer from high false positive rates due to the complexity of bacterial genomes and the need for different filtering cutoffs across sample types and sequencing depths. As datasets increase in size, the manual filtering required for high accuracy presents a significant obstacle. Here, we present AccuSNV, a novel deep learning-based tool for high-precision and automated bacterial SNV calling. Unlike traditional methods that process one sample at a time, AccuSNV leverages a convolutional neural network (CNN) that integrates alignment information across multiple samples, enhancing precision through learned across-sample patterns. We evaluated AccuSNV against seven popular SNV calling tools using simulated data from six bacterial species with varied sequencing depths, numbers of isolates, mutations, and divergence levels. To further validate its real-world utility, we tested AccuSNV on multiple curated bacterial datasets containing reported SNVs. In both simulated and real-world scenarios, AccuSNV consistently achieved the best performance. Moreover, AccuSNV provides comprehensive user-friendly downstream analysis modules and outputs, including mutation annotation information, phylogenetic inference, dN/dS calculations, and optional manual filtering. Together with the automated deep learning–based calling, these features make AccuSNV broadly accessible to users with different levels of computational expertise.
- Open Access
Phage-mediated lysis does not determine Cutibacterium acnes colonization on human skin
Despite Cutibacterium acnes being the most abundant and prevalent bacteria on human skin, only a single type of phage has been identified that infects this host. Here, we leverage this one-to-one system to systematically characterize how the phage-bacteria arms race shapes C. acnes evolution and community composition on individual people. Our analysis reveals a surprising lack of phage-mediated selection despite global prevalence of C. acnes phages. Analysis of anti-phage defense systems across 3,205 bacterial genomes revealed a limited, phylogenetically restricted defense repertoire under weak selective pressure to diversify or be maintained. Functional assays did not reveal alternative phage resistance mechanisms or fitness costs associated with defense gene carriage that could explain this limited immune arsenal. This lack of pressure to maintain phage resistance could not be explained by lack of phage colonization, as examination of 471 global human facial skin metagenomes demonstrated that even in samples with high virus-to-microbe ratio, phage-sensitive clades dominate on-person populations. Together, these findings indicate that phage pressure, while present, does not play a critical role in determining strain fitness and success within C. acnes populations on human skin. We propose that this observed weak phage-mediated selective pressure can be explained by the anatomy of skin: C. acnes growth is thought to occur at the bottom of pores, where exposure to phage may be limited by physical barriers. Together, this portrait of a static arms race provides a strong contrast with other microbial species in different ecosystems and expands understanding of phage-bacteria interactions in the human microbiome.
Open AccessPhage-mediated lysis does not determine Cutibacterium acnes colonization on human skin
Despite Cutibacterium acnes being the most abundant and prevalent bacteria on human skin, only a single type of phage has been identified that infects this host. Here, we leverage this one-to-one system to systematically characterize how the phage-bacteria arms race shapes C. acnes evolution and community composition on individual people. Our analysis reveals a surprising lack of phage-mediated selection despite global prevalence of C. acnes phages. Analysis of anti-phage defense systems across 3,205 bacterial genomes revealed a limited, phylogenetically restricted defense repertoire under weak selective pressure to diversify or be maintained. Functional assays did not reveal alternative phage resistance mechanisms or fitness costs associated with defense gene carriage that could explain this limited immune arsenal. This lack of pressure to maintain phage resistance could not be explained by lack of phage colonization, as examination of 471 global human facial skin metagenomes demonstrated that even in samples with high virus-to-microbe ratio, phage-sensitive clades dominate on-person populations. Together, these findings indicate that phage pressure, while present, does not play a critical role in determining strain fitness and success within C. acnes populations on human skin. We propose that this observed weak phage-mediated selective pressure can be explained by the anatomy of skin: C. acnes growth is thought to occur at the bottom of pores, where exposure to phage may be limited by physical barriers. Together, this portrait of a static arms race provides a strong contrast with other microbial species in different ecosystems and expands understanding of phage-bacteria interactions in the human microbiome.
- Open Access
- 12 August 2025
- Published by Elsevier in Cell Reports
- Vol. 44 (8) , 116134
Intraspecies associations from strain-rich metagenome samples
- 12 August 2025
- Published by Elsevier in Cell Reports
- Vol. 44 (8) , 116134
Open AccessIntraspecies associations from strain-rich metagenome samples
- Open Access
Intraspecies associations from strain-rich metagenome samples
Genetically distinct strains of a species can vary widely in phenotype, reducing the utility of species-resolved microbiome measurements for detecting associations with health or disease. While metagenomics theoretically provides information on all strains in a sample, current strain-resolved analysis methods face a tradeoff: de novo genotyping approaches can detect novel strains but struggle when applied to strain-rich or low-coverage samples, while reference database methods work robustly across sample types but are insensitive to novel diversity. We present PHLAME, a method that bridges this divide by combining the advantages of reference-based approaches with novelty awareness. PHLAME explicitly defines clades at multiple phylogenetic levels and introduces a probabilistic, mutation-based, framework to accurately quantify novelty from the nearest reference. By applying PHLAME to publicly available human skin and vaginal metagenomes, we uncover previously undetected clade associations with coexisting species, geography, and host age. The ability to characterize intraspecies associations and dynamics in previously inaccessible environments will propel new mechanistic insights from accumulating metagenomic data.
Open AccessIntraspecies associations from strain-rich metagenome samples
Genetically distinct strains of a species can vary widely in phenotype, reducing the utility of species-resolved microbiome measurements for detecting associations with health or disease. While metagenomics theoretically provides information on all strains in a sample, current strain-resolved analysis methods face a tradeoff: de novo genotyping approaches can detect novel strains but struggle when applied to strain-rich or low-coverage samples, while reference database methods work robustly across sample types but are insensitive to novel diversity. We present PHLAME, a method that bridges this divide by combining the advantages of reference-based approaches with novelty awareness. PHLAME explicitly defines clades at multiple phylogenetic levels and introduces a probabilistic, mutation-based, framework to accurately quantify novelty from the nearest reference. By applying PHLAME to publicly available human skin and vaginal metagenomes, we uncover previously undetected clade associations with coexisting species, geography, and host age. The ability to characterize intraspecies associations and dynamics in previously inaccessible environments will propel new mechanistic insights from accumulating metagenomic data.
- Open Access
Intraspecies warfare restricts strain coexistence in human skin microbiomes
Determining why only a fraction of encountered or applied strains engraft in a given person’s microbiome is crucial for understanding and engineering these communities. Previous work has established that metabolic competition can restrict colonization success in vivo , but the relevance of bacterial warfare in preventing commensal engraftment has been less explored. Here, we demonstrate that intraspecies warfare presents a significant barrier to strain coexistence in the human skin microbiome by profiling 14,884 pairwise interactions between Staphylococcus epidermidis isolates cultured from eighteen people from six families. We find that intraspecies antagonisms are abundant, mechanistically diverse, independent of strain relatedness, and consistent with rapid evolution via horizontal gene transfer. Critically, these antagonisms are significantly depleted among strains residing on the same person relative to random assemblages, indicating a significant in vivo role. Together, our results emphasize that accounting for intraspecies warfare may be essential to the design of long-lasting probiotic therapeutics.
Open AccessIntraspecies warfare restricts strain coexistence in human skin microbiomes
Determining why only a fraction of encountered or applied strains engraft in a given person’s microbiome is crucial for understanding and engineering these communities. Previous work has established that metabolic competition can restrict colonization success in vivo , but the relevance of bacterial warfare in preventing commensal engraftment has been less explored. Here, we demonstrate that intraspecies warfare presents a significant barrier to strain coexistence in the human skin microbiome by profiling 14,884 pairwise interactions between Staphylococcus epidermidis isolates cultured from eighteen people from six families. We find that intraspecies antagonisms are abundant, mechanistically diverse, independent of strain relatedness, and consistent with rapid evolution via horizontal gene transfer. Critically, these antagonisms are significantly depleted among strains residing on the same person relative to random assemblages, indicating a significant in vivo role. Together, our results emphasize that accounting for intraspecies warfare may be essential to the design of long-lasting probiotic therapeutics.
- Open Access
Previously hidden intraspecies dynamics underlie the apparent stability of two important skin microbiome species
Adult human facial skin microbiomes are remarkably similar at the species-level, dominated by Cutibacterium acnes and Staphylococcus epidermidis, yet each person harbors a unique community of strains. Understanding how person-specific communities assemble is critical for designing microbiome-based therapies. Here, using 4,055 isolate genomes and 360 metagenomes, we reconstruct on-person evolutionary history to reveal on and between-person strain dynamics. We find that multiple cells are typically involved in transmission, indicating ample opportunity for migration. Despite this accessibility, family members share only some of their strains. S. epidermidis communities are dynamic, with each strain persisting for an average of only 2 years. C. acnes strains are more stable and have a higher colonization rate during the transition to an adult facial skin microbiome, suggesting this window could facilitate engraftment of therapeutic strains. These previously undetectable dynamics may influence the design of microbiome therapeutics and motivate the study of their effects on hosts.
Open AccessPreviously hidden intraspecies dynamics underlie the apparent stability of two important skin microbiome species
Adult human facial skin microbiomes are remarkably similar at the species-level, dominated by Cutibacterium acnes and Staphylococcus epidermidis, yet each person harbors a unique community of strains. Understanding how person-specific communities assemble is critical for designing microbiome-based therapies. Here, using 4,055 isolate genomes and 360 metagenomes, we reconstruct on-person evolutionary history to reveal on and between-person strain dynamics. We find that multiple cells are typically involved in transmission, indicating ample opportunity for migration. Despite this accessibility, family members share only some of their strains. S. epidermidis communities are dynamic, with each strain persisting for an average of only 2 years. C. acnes strains are more stable and have a higher colonization rate during the transition to an adult facial skin microbiome, suggesting this window could facilitate engraftment of therapeutic strains. These previously undetectable dynamics may influence the design of microbiome therapeutics and motivate the study of their effects on hosts.