Bruce Rannala
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Bruce Rannala has a total of 55 co-authors. Here are the top co-authors by number of shared publications:
- Ziheng Yang
Department of Genetics, Evolution, and Environment, University College London, London WC1E 6BT, UK
13 shared publications - Anna Audrey Nagel
Washington University in St. Louis, St. Louis One Brookings Drive, Department of Biology MO 63130
5 shared publications Xiyun Jiao
Department of Statistics and Data Science, China Southern University of Science and Technology, Shenzhen 518055, China
5 shared publicationsZiheng Yang
Department of Genetics, University College London, Gower Street, London WC1E 6BT, UK
5 shared publications- Tomáš Flouri
Department of Genetics, Evolution, and Environment, University College London , Gower Street, London WC1E 6BT
3 shared publications Ying Wang
Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing 100101, China
3 shared publications
Latest Publications
- 18 July 2026
- Published by Springer Nature in Conservation Genetics Resources
- Vol. 18 (3) , 33
Mongrail 2.0: Bayesian inference of hybrids using population genomic data
- 18 July 2026
- Published by Springer Nature in Conservation Genetics Resources
- Vol. 18 (3) , 33
Mongrail 2.0: Bayesian inference of hybrids using population genomic data
- 16 July 2026
- Published by Springer Nature in Heredity
- Vol. 135 (9) , 660-673
Improved Bayesian inference of hybrids using genome sequences
A Bayesian hybrid inference method is developed which infers hybrids and backcrosses across two generations using sampled genomes from two populations. The method improves on that of Chakraborty and Rannala (2023) by accounting for uncertainty of population haplotype frequencies and correctly marginalizing over haplotypes while still modeling linkage and recombination across the genome. In analyses of simulated data the new method produced posterior probabilities nearly identical to the method of Chakraborty and Rannala (2023) when sample sizes were large. For small sample sizes, posterior probabilities produced by the new method tended to be lower as expected since it accounts for additional uncertainties of population haplotype frequencies. Statistical performance of the new method as measured by the ROC (Receiver Operating Characteristic) curve appears equivalent to that of Chakraborty and Rannala (2023). The new method is applied to three recently published datasets for populations of kiwifruit (genus Actinidia), plateau fence lizard (Sceloporus tristichus) and puma (Puma concolor).
- 16 July 2026
- Published by Springer Nature in Heredity
- Vol. 135 (9) , 660-673
Improved Bayesian inference of hybrids using genome sequences
A Bayesian hybrid inference method is developed which infers hybrids and backcrosses across two generations using sampled genomes from two populations. The method improves on that of Chakraborty and Rannala (2023) by accounting for uncertainty of population haplotype frequencies and correctly marginalizing over haplotypes while still modeling linkage and recombination across the genome. In analyses of simulated data the new method produced posterior probabilities nearly identical to the method of Chakraborty and Rannala (2023) when sample sizes were large. For small sample sizes, posterior probabilities produced by the new method tended to be lower as expected since it accounts for additional uncertainties of population haplotype frequencies. Statistical performance of the new method as measured by the ROC (Receiver Operating Characteristic) curve appears equivalent to that of Chakraborty and Rannala (2023). The new method is applied to three recently published datasets for populations of kiwifruit (genus Actinidia), plateau fence lizard (Sceloporus tristichus) and puma (Puma concolor).
- Open Access
- 8 April 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Scaling up Bayesian population phylogenomics through virtual dimension reduction
Population phylogenomics uses sampled genomes to jointly infer population genetic processes (ancestral and contemporary population sizes, historical gene flow) and a phylogenetic tree relating species or populations including species split times. This challenging problem has been tackled most successfully in the Bayesian framework under the multispecies coalescent (MSC) model via Markov chain Monte Carlo (MCMC) computational algorithms. However, MCMC methods suffer from two serious problems: (i) mixing difficulties due to the high-dimensional state space with complex constraints, and (ii) the intrinsically serial nature of MCMC algorithms that defies parallelisation. To deal with both issues, we develop a new method, called Virtual Dimension Reduction allowing Parallelisation (VDRoP), that achieves the same MCMC mixing efficiency as dimension reduction through analytical integration of parameters, but without sacrificing parallel computation and without the restriction to conjugate priors. We implement the new method in the Bayesian program BPP and apply it to genomic datasets from Adansonia baobab trees, Anopheles mosquitoes, and Heliconius butterflies. The new algorithms reduce the run-time of MCMC analyses by 3 to 8 fold and improve the mixing efficiency by up to 50 fold for representative empirical datasets.
- 8 April 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessScaling up Bayesian population phylogenomics through virtual dimension reduction
Population phylogenomics uses sampled genomes to jointly infer population genetic processes (ancestral and contemporary population sizes, historical gene flow) and a phylogenetic tree relating species or populations including species split times. This challenging problem has been tackled most successfully in the Bayesian framework under the multispecies coalescent (MSC) model via Markov chain Monte Carlo (MCMC) computational algorithms. However, MCMC methods suffer from two serious problems: (i) mixing difficulties due to the high-dimensional state space with complex constraints, and (ii) the intrinsically serial nature of MCMC algorithms that defies parallelisation. To deal with both issues, we develop a new method, called Virtual Dimension Reduction allowing Parallelisation (VDRoP), that achieves the same MCMC mixing efficiency as dimension reduction through analytical integration of parameters, but without sacrificing parallel computation and without the restriction to conjugate priors. We implement the new method in the Bayesian program BPP and apply it to genomic datasets from Adansonia baobab trees, Anopheles mosquitoes, and Heliconius butterflies. The new algorithms reduce the run-time of MCMC analyses by 3 to 8 fold and improve the mixing efficiency by up to 50 fold for representative empirical datasets.
- Open Access
Improved Bayesian inference of hybrids using genome sequences
A Bayesian hybrid inference method is developed which infers hybrids and backcrosses across two generations using sampled genomes from two populations. The method improves on that of Chakraborty and Rannala (2023) by accounting for uncertainty of population haplotype frequencies and correctly marginalizing over haplotypes while still modeling linkage and recombination across the genome. In analyses of simulated data the new method produced posterior probabilities nearly identical to the method of Chakraborty and Rannala (2023) when sample sizes were large. For small sample sizes, posterior probabilities produced by the new method tended to be lower as expected since it accounts for additional uncertainties of population haplotype frequencies. Statistical performance of the new method as measured by the ROC (Receiver Operation Characteristic) curve, appears equivalent to that of Chakraborty and Rannala (2023). The new method is applied to three recently published datasets for populations of kiwifruit (genus Actinidia ), plateau fence lizard ( Sceloporus tristichus ) and puma ( Puma concolor ).
Open AccessImproved Bayesian inference of hybrids using genome sequences
A Bayesian hybrid inference method is developed which infers hybrids and backcrosses across two generations using sampled genomes from two populations. The method improves on that of Chakraborty and Rannala (2023) by accounting for uncertainty of population haplotype frequencies and correctly marginalizing over haplotypes while still modeling linkage and recombination across the genome. In analyses of simulated data the new method produced posterior probabilities nearly identical to the method of Chakraborty and Rannala (2023) when sample sizes were large. For small sample sizes, posterior probabilities produced by the new method tended to be lower as expected since it accounts for additional uncertainties of population haplotype frequencies. Statistical performance of the new method as measured by the ROC (Receiver Operation Characteristic) curve, appears equivalent to that of Chakraborty and Rannala (2023). The new method is applied to three recently published datasets for populations of kiwifruit (genus Actinidia ), plateau fence lizard ( Sceloporus tristichus ) and puma ( Puma concolor ).
- Open Access
- 12 December 2025
- Published by Oxford University Press (OUP) in Molecular Biology and Evolution
- Vol. 43 (1)
On the robustness of Bayesian inference of gene flow to intragenic recombination and natural selection
The multispecies coalescent (MSC) model provides a framework for detecting gene flow using genomic data, including between sister species. However, the robustness of the inference to violations of model assumptions are poorly understood. Here, we use simulation to study the false positive rate of a Bayesian test of gene flow under the MSC with multiple influencing factors including recombination, natural selection, discrete versus continuous gene flow, variable species divergence time, and gene flow involving sister versus nonsister lineages. We find that in almost all scenarios examined the test has very low false positives. However, the test of gene flow between sister lineages may be prone to high false positives in cases of very recent species divergence and very high recombination rate. At low recombination rates, the test is robust to selective sweeps, background selection and balancing selection, although prolonged balancing selection can lead to false signals of gene flow between sister lineages. The impact of excessive recombination on the test of gene flow between sisters may be assessed by using a smaller number of sequences for each species and by considering shorter sequences at each locus. Recent species divergence alone (with no recombination) does not cause false positives in tests of gene flow, contrary to previous claims. The test of gene flow between nonsister lineages is robust to recombination at all divergence levels. Our findings provide guidance for reliable inference of gene flow using coalescent methods and highlight the need for care in conducting and interpreting simulation experiments.
- 12 December 2025
- Published by Oxford University Press (OUP) in Molecular Biology and Evolution
- Vol. 43 (1)
Open AccessOn the robustness of Bayesian inference of gene flow to intragenic recombination and natural selection
The multispecies coalescent (MSC) model provides a framework for detecting gene flow using genomic data, including between sister species. However, the robustness of the inference to violations of model assumptions are poorly understood. Here, we use simulation to study the false positive rate of a Bayesian test of gene flow under the MSC with multiple influencing factors including recombination, natural selection, discrete versus continuous gene flow, variable species divergence time, and gene flow involving sister versus nonsister lineages. We find that in almost all scenarios examined the test has very low false positives. However, the test of gene flow between sister lineages may be prone to high false positives in cases of very recent species divergence and very high recombination rate. At low recombination rates, the test is robust to selective sweeps, background selection and balancing selection, although prolonged balancing selection can lead to false signals of gene flow between sister lineages. The impact of excessive recombination on the test of gene flow between sisters may be assessed by using a smaller number of sequences for each species and by considering shorter sequences at each locus. Recent species divergence alone (with no recombination) does not cause false positives in tests of gene flow, contrary to previous claims. The test of gene flow between nonsister lineages is robust to recombination at all divergence levels. Our findings provide guidance for reliable inference of gene flow using coalescent methods and highlight the need for care in conducting and interpreting simulation experiments.
- 23 September 2025
- Published by Oxford University Press (OUP) in Genetics
- Vol. 232 (1)
Mutation ages and population origins inferred from genomes in structured populations
Inferring the time of origin (age) of mutations is an old question in population genetics and inferring their population of origin has become of particular interest with the sequencing of the Neanderthal genome. However, existing methods to infer mutation ages and populations of origin do not explicitly consider population structure, migration rates, and divergence times, which may bias estimates, and it is unclear how to even apply single-population estimators to structured populations. We develop a method to jointly estimate the time and population of origin of a mutation (as well as the ancestral and derived states) in a structured population using population genomic data and examine its statistical performance using simulations. Results indicate that mutation age and population of origin can be quite uncertain, even with long sequences or many samples, but this uncertainty is accurately captured using credible intervals/sets. The ancestral nucleotide state is relatively easy to infer. We apply our method to whole genome data from the 1,000 Genomes Project, analyzing 7 SNP mutations from 6 genes associated with human skin pigmentation for populations from Great Britain, China, and Kenya. Our results partially support previous conclusions, with the putative ancestral alleles from the literature matching our inferences, while the mutation age estimates only overlap in some cases.
- 23 September 2025
- Published by Oxford University Press (OUP) in Genetics
- Vol. 232 (1)
Mutation ages and population origins inferred from genomes in structured populations
Inferring the time of origin (age) of mutations is an old question in population genetics and inferring their population of origin has become of particular interest with the sequencing of the Neanderthal genome. However, existing methods to infer mutation ages and populations of origin do not explicitly consider population structure, migration rates, and divergence times, which may bias estimates, and it is unclear how to even apply single-population estimators to structured populations. We develop a method to jointly estimate the time and population of origin of a mutation (as well as the ancestral and derived states) in a structured population using population genomic data and examine its statistical performance using simulations. Results indicate that mutation age and population of origin can be quite uncertain, even with long sequences or many samples, but this uncertainty is accurately captured using credible intervals/sets. The ancestral nucleotide state is relatively easy to infer. We apply our method to whole genome data from the 1,000 Genomes Project, analyzing 7 SNP mutations from 6 genes associated with human skin pigmentation for populations from Great Britain, China, and Kenya. Our results partially support previous conclusions, with the putative ancestral alleles from the literature matching our inferences, while the mutation age estimates only overlap in some cases.
- Open Access
- 13 February 2025
- Published by The Royal Society in Philosophical Transactions of the Royal Society B: Biological Sciences
- Vol. 380 (1919) , 20230309
Reading tree leaves: inferring speciation and extinction processes using phylogenies
The birth–death process (BDP) is widely used in evolutionary biology as a model for generating phylogenetic trees of species. The generalized birth–death process (GBDP) allows rate variation over time, with speciation and extinction rates to be arbitrary functions of time. Here we review the probability theory underpinning the GBDP as a model of cladogenesis and recent findings concerning its identifiability. The GBDP with arbitrary continuous rate functions has been shown to be non-identifiable from lineage-through-time data: even with species phylogenies of infinite size the parameters cannot be estimated. However, a restricted class of BDPs with piecewise-constant rates has been shown to be identifiable. We review and illustrate these results using simple examples and discuss their implications for biologists interested in inferring the past tempo and mode of evolution using reconstructed phylogenetic trees. This article is part of the theme issue ‘“A mathematical theory of evolution”: phylogenetic models dating back 100 years’.
- 13 February 2025
- Published by The Royal Society in Philosophical Transactions of the Royal Society B: Biological Sciences
- Vol. 380 (1919) , 20230309
Open AccessReading tree leaves: inferring speciation and extinction processes using phylogenies
The birth–death process (BDP) is widely used in evolutionary biology as a model for generating phylogenetic trees of species. The generalized birth–death process (GBDP) allows rate variation over time, with speciation and extinction rates to be arbitrary functions of time. Here we review the probability theory underpinning the GBDP as a model of cladogenesis and recent findings concerning its identifiability. The GBDP with arbitrary continuous rate functions has been shown to be non-identifiable from lineage-through-time data: even with species phylogenies of infinite size the parameters cannot be estimated. However, a restricted class of BDPs with piecewise-constant rates has been shown to be identifiable. We review and illustrate these results using simple examples and discuss their implications for biologists interested in inferring the past tempo and mode of evolution using reconstructed phylogenetic trees. This article is part of the theme issue ‘“A mathematical theory of evolution”: phylogenetic models dating back 100 years’.
- Open Access
Mutation ages and population origins inferred from genomes in structured populations
Inferring the time of origin (age) of mutations is an old question in population genetics and inferring their population of origin has become of particular interest with the sequencing of the Neanderthal genome. However, existing methods to infer mutation ages and populations of origin do not explicitly consider population structure, migration rates, and divergence times, which may bias estimates, and it is unclear how to even apply single-population estimators to structured populations. We develop a method to jointly estimate the time and population of origin of a mutation (as well as the ancestral and derived states) in a structured population using population genomic data and examine its statistical performance using simulations. Results indicate that mutation age and population of origin can be quite uncertain, even with long sequences or many samples, but this uncertainty is accurately captured using credible intervals/sets. The ancestral nucleotide state is relatively easy to infer. We apply our method to whole genome data from the 1000 Genomes Project, analyzing seven SNP mutations from six genes associated with human skin pigmentation for populations from Great Britain, China, and Kenya. Our results partially support previous conclusions, with the putative ancestral alleles from the literature matching our inferences, while the mutation age estimates only overlap in some cases. Furthermore, there are non-trivial posterior probabilities of recurrence for three of the mutations.
Open AccessMutation ages and population origins inferred from genomes in structured populations
Inferring the time of origin (age) of mutations is an old question in population genetics and inferring their population of origin has become of particular interest with the sequencing of the Neanderthal genome. However, existing methods to infer mutation ages and populations of origin do not explicitly consider population structure, migration rates, and divergence times, which may bias estimates, and it is unclear how to even apply single-population estimators to structured populations. We develop a method to jointly estimate the time and population of origin of a mutation (as well as the ancestral and derived states) in a structured population using population genomic data and examine its statistical performance using simulations. Results indicate that mutation age and population of origin can be quite uncertain, even with long sequences or many samples, but this uncertainty is accurately captured using credible intervals/sets. The ancestral nucleotide state is relatively easy to infer. We apply our method to whole genome data from the 1000 Genomes Project, analyzing seven SNP mutations from six genes associated with human skin pigmentation for populations from Great Britain, China, and Kenya. Our results partially support previous conclusions, with the putative ancestral alleles from the literature matching our inferences, while the mutation age estimates only overlap in some cases. Furthermore, there are non-trivial posterior probabilities of recurrence for three of the mutations.
- Open Access
- 13 January 2025
- Published by Oxford University Press (OUP) in Evolutionary Journal of the Linnean Society
- Vol. 4 (1) , kzaf016
Recombination and phylogenetic inference
I explore the problem of inferring phylogenetic trees in the presence of recombination. Two widely used approaches are considered: concatenation methods assume all loci have one underlying gene tree; and species tree inference methods assume one gene tree underlies each locus (no intralocus recombination) and loci have independent gene trees (high interlocus recombination). The impact of recombination is different under these two approaches. Three strategies for addressing the impacts of recombination are considered: (i) studies of the statistical robustness of phylogenetic inference methods when recombination occurs and is not accounted for (if impacts are minimal, recombination can be safely ignored); (ii) methods that accommodate recombination by identifying recombinant regions to either eliminate recombinant loci (to reduce intralocus recombination) or to choose loci that are separated by multiple recombinations (to increase interlocus recombination); and (iii) methods for phylogenetic inference that aim to accommodate recombination by inferring breakpoints between regions of sequences with different gene trees or allow varying topology along a sequence. I conclude that recombination is likely to be more detrimental for concatenation methods, having little impact on topology or divergence time estimates for species tree inference methods. Recombination detection may not be necessary when performing species tree inference, and eliminating recombinant loci may bias parameter estimates. Methods allowing gene trees to vary across the genome still lack theory-based criteria for combining inferred gene trees to estimate a species tree; this could in principle be done using a multispecies coalescent model with recombination but is a considerable technical challenge.
- 13 January 2025
- Published by Oxford University Press (OUP) in Evolutionary Journal of the Linnean Society
- Vol. 4 (1) , kzaf016
Open AccessRecombination and phylogenetic inference
I explore the problem of inferring phylogenetic trees in the presence of recombination. Two widely used approaches are considered: concatenation methods assume all loci have one underlying gene tree; and species tree inference methods assume one gene tree underlies each locus (no intralocus recombination) and loci have independent gene trees (high interlocus recombination). The impact of recombination is different under these two approaches. Three strategies for addressing the impacts of recombination are considered: (i) studies of the statistical robustness of phylogenetic inference methods when recombination occurs and is not accounted for (if impacts are minimal, recombination can be safely ignored); (ii) methods that accommodate recombination by identifying recombinant regions to either eliminate recombinant loci (to reduce intralocus recombination) or to choose loci that are separated by multiple recombinations (to increase interlocus recombination); and (iii) methods for phylogenetic inference that aim to accommodate recombination by inferring breakpoints between regions of sequences with different gene trees or allow varying topology along a sequence. I conclude that recombination is likely to be more detrimental for concatenation methods, having little impact on topology or divergence time estimates for species tree inference methods. Recombination detection may not be necessary when performing species tree inference, and eliminating recombinant loci may bias parameter estimates. Methods allowing gene trees to vary across the genome still lack theory-based criteria for combining inferred gene trees to estimate a species tree; this could in principle be done using a multispecies coalescent model with recombination but is a considerable technical challenge.
- Open Access
- 30 July 2024
- Published by Oxford University Press (OUP) in Systematic Biology
- Vol. 73 (6) , 964-978
Bayesian Inference Under the Multispecies Coalescent with Ancient DNA Sequences
Ancient DNA (aDNA) is increasingly being used to investigate questions such as the phylogenetic relationships and divergence times of extant and extinct species. If aDNA samples are sufficiently old, expected branch lengths (in units of nucleotide substitutions) are reduced relative to contemporary samples. This can be accounted for by incorporating sample ages into phylogenetic analyses. Existing methods that use tip (sample) dates infer gene trees rather than species trees, which can lead to incorrect or biased inferences of the species tree. Methods using a multispecies coalescent (MSC) model overcome these issues. We developed an MSC model with tip dates and implemented it in the program BPP. The method performed well for a range of biologically realistic scenarios, estimating calibrated divergence times and mutation rates precisely. Simulations suggest that estimation precision can be best improved by prioritizing sampling of many loci and more ancient samples. Incorrectly treating ancient samples as contemporary in analyzing simulated data, mimicking a common practice of empirical analyses, led to large systematic biases in model parameters, including divergence times. Two genomic datasets of mammoths and elephants were analyzed, demonstrating the method’s empirical utility.
- 30 July 2024
- Published by Oxford University Press (OUP) in Systematic Biology
- Vol. 73 (6) , 964-978
Open AccessBayesian Inference Under the Multispecies Coalescent with Ancient DNA Sequences
Ancient DNA (aDNA) is increasingly being used to investigate questions such as the phylogenetic relationships and divergence times of extant and extinct species. If aDNA samples are sufficiently old, expected branch lengths (in units of nucleotide substitutions) are reduced relative to contemporary samples. This can be accounted for by incorporating sample ages into phylogenetic analyses. Existing methods that use tip (sample) dates infer gene trees rather than species trees, which can lead to incorrect or biased inferences of the species tree. Methods using a multispecies coalescent (MSC) model overcome these issues. We developed an MSC model with tip dates and implemented it in the program BPP. The method performed well for a range of biologically realistic scenarios, estimating calibrated divergence times and mutation rates precisely. Simulations suggest that estimation precision can be best improved by prioritizing sampling of many loci and more ancient samples. Incorrectly treating ancient samples as contemporary in analyzing simulated data, mimicking a common practice of empirical analyses, led to large systematic biases in model parameters, including divergence times. Two genomic datasets of mammoths and elephants were analyzed, demonstrating the method’s empirical utility.