Alexandros Stamatakis

Publications of Alexandros Stamatakis

• A functional phylogenomic view of the seed plants.

  Authors: Ernest K Lee, Angelica Cibrian-Jaramillo, Sergios-Orestis Kolokotronis, Manpreet S Katari, Alexandros Stamatakis, Michael Ott, Joanna C Chiu, Damon P Little, Dennis Wm Stevenson, W Richard McCombie, Robert A Martienssen, Gloria Coruzzi, Rob Desalle

  PLoS genetics. 12/2011; 7(12):e1002411.

  A novel result of the current research is the development and implementation of a unique functional phylogenomic approach that explores the genomic origins of seed plant diversification. We first useA novel result of the current research is the development and implementation of a unique functional phylogenomic approach that explores the genomic origins of seed plant diversification. We first use 22,833 sets of orthologs from the nuclear genomes of 101 genera across land plants to reconstruct their phylogenetic relationships. One of the more salient results is the resolution of some enigmatic relationships in seed plant phylogeny, such as the placement of Gnetales as sister to the rest of the gymnosperms. In using this novel phylogenomic approach, we were also able to identify overrepresented functional gene ontology categories in genes that provide positive branch support for major nodes prompting new hypotheses for genes associated with the diversification of angiosperms. For example, RNA interference (RNAi) has played a significant role in the divergence of monocots from other angiosperms, which has experimental support in Arabidopsis and rice. This analysis also implied that the second largest subunit of RNA polymerase IV and V (NRPD2) played a prominent role in the divergence of gymnosperms. This hypothesis is supported by the lack of 24nt siRNA in conifers, the maternal control of small RNA in the seeds of flowering plants, and the emergence of double fertilization in angiosperms. Our approach takes advantage of genomic data to define orthologs, reconstruct relationships, and narrow down candidate genes involved in plant evolution within a phylogenomic view of species' diversification.

  Download

• Uncovering hidden phylogenetic consensus in large data sets.

  Authors: Nicholas D Pattengale, Andre J Aberer, Krister M Swenson, Alexandros Stamatakis, Bernard M E Moret

  IEEE/ACM transactions on computational biology and bioinformatics / IEEE, ACM. 02/2011; 8(4):902-11.

  Many of the steps in phylogenetic reconstruction can be confounded by “rogue” taxa—taxa that cannot be placed with assurance anywhere within the tree, indeed, whose location within the tree variesMany of the steps in phylogenetic reconstruction can be confounded by “rogue” taxa—taxa that cannot be placed with assurance anywhere within the tree, indeed, whose location within the tree varies with almost any choice of algorithm or parameters. Phylogenetic consensus methods, in particular, are known to suffer from this problem. In this paper, we provide a novel framework to define and identify rogue taxa. In this framework, we formulate a bicriterion optimization problem, the relative information criterion, that models the net increase in useful information present in the consensus tree when certain taxa are removed from the input data. We also provide an effective greedy heuristic to identify a subset of rogue taxa and use this heuristic in a series of experiments, with both pathological examples from the literature and a collection of large biological data sets. As the presence of rogue taxa in a set of bootstrap replicates can lead to deceivingly poor support values, we propose a procedure to recompute support values in light of the rogue taxa identified by our algorithm; applying this procedure to our biological data sets caused a large number of edges to move from “unsupported” to “supported” status, indicating that many existing phylogenies should be recomputed and reevaluated to reduce any inaccuracies introduced by rogue taxa. We also discuss the implementation issues encountered while integrating our algorithm into RAxML v7.2.7, particularly those dealing with scaling up the analyses. This integration enables practitioners to benefit from our algorithm in the analysis of very large data sets (up to 2,500 taxa and 10,000 trees, although we present the results of even larger analyses).

• FPGA Optimizations for a Pipelined Floating-Point Exponential Unit.

  Authors: Nikolaos Alachiotis, Alexandros Stamatakis

  Reconfigurable Computing: Architectures, Tools and Applications - 7th International Symposium, ARC 2011, Belfast, UK, March 23-25, 2011. Proceedings; 01/2011

• FPGA Acceleration of the Phylogenetic Parsimony Kernel?

  Authors: Nikolaos Alachiotis, Alexandros Stamatakis

  International Conference on Field Programmable Logic and Applications, FPL 2011, September 5-7, Chania, Crete, Greece; 01/2011

• Uncovering Hidden Phylogenetic Consensus in Large Data Sets.

  Authors: Nicholas D. Pattengale, Andre J. Aberer, Krister M. Swenson, Alexandros Stamatakis, Bernard M. E. Moret

  IEEE/ACM Trans. Comput. Biology Bioinform. 01/2011; 8:902-911.

• Quantifying the phylodynamic forces driving papillomavirus evolution.

  Authors: Marc Gottschling, Markus Göker, Alexandros Stamatakis, Olaf R P Bininda-Emonds, Ingo Nindl, Ignacio G Bravo

  Molecular biology and evolution. 01/2011; 28(7):2101-13.

  The associations between pathogens and their hosts are complex and can result from a variety of evolutionary processes including codivergence, lateral transfer, or duplication. Papillomaviruses (PVs)The associations between pathogens and their hosts are complex and can result from a variety of evolutionary processes including codivergence, lateral transfer, or duplication. Papillomaviruses (PVs) are double-stranded DNA viruses ubiquitously present in mammals and are a suitable target for rigorous statistical tests of potential virus-host codivergence. We analyze the evolutionary dynamics of PV diversification by comparing robust phylogenies of PVs and their respective hosts using different statistical approaches to assess topological and branch-length congruence. Mammalian PVs segregated into four diverse major clades that overlapped to varying degrees in terms of their mammalian host lineages. The hypothesis that PVs and hosts evolved independently was globally rejected (P = 0.0001), although only 90 of 207 virus-host associations (43%) were significant in individual tests. Virus-host codivergence accounted roughly for one-third of the evolutionary events required to reconcile PV-host evolutionary histories. When virus-host associations were analyzed locally within each of the four viral clades, numerous independent topological congruencies were identified that were incompatible with respect to the global trees. These results support an evolutionary scenario in which early PV radiation was followed by independent codivergence between viruses within each of the major clades and their hosts. Moreover, heterogeneous groups of closely related PVs infecting non-related hosts suggest several interspecies transmission events. Our results argue thus for the importance of alternative events in PV evolution, in contrast to the prevailing opinion that these viruses show a high degree of host specificity and codivergence.

• Time and memory efficient likelihood-based tree searches on phylogenomic alignments with missing data.

  Authors: Alexandros Stamatakis, Nikolaos Alachiotis

  Bioinformatics (Oxford, England). 06/2010; 26(12):i132-9.

  The current molecular data explosion poses new challenges for large-scale phylogenomic analyses that can comprise hundreds or even thousands of genes. A property that characterizes phylogenomicThe current molecular data explosion poses new challenges for large-scale phylogenomic analyses that can comprise hundreds or even thousands of genes. A property that characterizes phylogenomic datasets is that they tend to be gappy, i.e. can contain taxa with (many and disparate) missing genes. In current phylogenomic analyses, this type of alignment gappyness that is induced by missing data frequently exceeds 90%. We present and implement a generally applicable mechanism that allows for reducing memory footprints of likelihood-based [maximum likelihood (ML) or Bayesian] phylogenomic analyses proportional to the amount of missing data in the alignment. We also introduce a set of algorithmic rules to efficiently conduct tree searches via subtree pruning and re-grafting moves using this mechanism. On a large phylogenomic DNA dataset with 2177 taxa, 68 genes and a gappyness of 90%, we achieve a memory footprint reduction from 9 GB down to 1 GB, a speedup for optimizing ML model parameters of 11, and accelerate the Subtree Pruning Regrafting tree search phase by factor 16. Thus, our approach can be deployed to improve efficiency for the two most important resources, CPU time and memory, by up to one order of magnitude. Current open-source version of RAxML v7.2.6 available at http://wwwkramer.in.tum.de/exelixis/software.html.

• How many bootstrap replicates are necessary?

  Authors: Nicholas D Pattengale, Masoud Alipour, Olaf R P Bininda-Emonds, Bernard M E Moret, Alexandros Stamatakis

  Journal of computational biology : a journal of computational molecular cell biology. 03/2010; 17(3):337-54.

  Phylogenetic bootstrapping (BS) is a standard technique for inferring confidence values on phylogenetic trees that is based on reconstructing many trees from minor variations of the input data, treesPhylogenetic bootstrapping (BS) is a standard technique for inferring confidence values on phylogenetic trees that is based on reconstructing many trees from minor variations of the input data, trees called replicates. BS is used with all phylogenetic reconstruction approaches, but we focus here on one of the most popular, maximum likelihood (ML). Because ML inference is so computationally demanding, it has proved too expensive to date to assess the impact of the number of replicates used in BS on the relative accuracy of the support values. For the same reason, a rather small number (typically 100) of BS replicates are computed in real-world studies. Stamatakis et al. recently introduced a BS algorithm that is 1 to 2 orders of magnitude faster than previous techniques, while yielding qualitatively comparable support values, making an experimental study possible. In this article, we propose stopping criteria--that is, thresholds computed at runtime to determine when enough replicates have been generated--and we report on the first large-scale experimental study to assess the effect of the number of replicates on the quality of support values, including the performance of our proposed criteria. We run our tests on 17 diverse real-world DNA--single-gene as well as multi-gene--datasets, which include 125-2,554 taxa. We find that our stopping criteria typically stop computations after 100-500 replicates (although the most conservative criterion may continue for several thousand replicates) while producing support values that correlate at better than 99.5% with the reference values on the best ML trees. Significantly, we also find that the stopping criteria can recommend very different numbers of replicates for different datasets of comparable sizes. Our results are thus twofold: (i) they give the first experimental assessment of the effect of the number of BS replicates on the quality of support values returned through BS, and (ii) they validate our proposals for stopping criteria. Practitioners will no longer have to enter a guess nor worry about the quality of support values; moreover, with most counts of replicates in the 100-500 range, robust BS under ML inference becomes computationally practical for most datasets. The complete test suite is available at http://lcbb.epfl.ch/BS.tar.bz2, and BS with our stopping criteria is included in the latest release of RAxML v7.2.5, available at http://wwwkramer.in.tum.de/exelixis/software.html.

• Parallel computation of phylogenetic consensus trees.

  Authors: Andre J. Aberer, Nicholas D. Pattengale, Alexandros Stamatakis

  Procedia CS. 01/2010; 1:1065-1073.

• Parallelized phylogenetic post-analysis on multi-core architectures.

  Authors: Andre J. Aberer, Nicholas D. Pattengale, Alexandros Stamatakis

  J. Comput. Science. 01/2010; 1:107-114.

• Efficient floating-point logarithm unit for FPGAs.

  Authors: Nikolaos Alachiotis, Alexandros Stamatakis

  24th IEEE International Symposium on Parallel and Distributed Processing, IPDPS 2010, Atlanta, Georgia, USA, 19-23 April 2010 - Workshop Proceedings; 01/2010

• Maximum Likelihood Analyses of 3,490 rbcL Sequences: Scalability of Comprehensive Inference versus Group-Specific Taxon Sampling.

  Authors: Alexandros Stamatakis, Markus Göker, Guido W Grimm

  Evolutionary bioinformatics online. 01/2010; 6:73-90.

  The constant accumulation of sequence data poses new computational and methodological challenges for phylogenetic inference, since multiple sequence alignments grow both in the horizontal (number ofThe constant accumulation of sequence data poses new computational and methodological challenges for phylogenetic inference, since multiple sequence alignments grow both in the horizontal (number of base pairs, phylogenomic alignments) as well as vertical (number of taxa) dimension. Put aside the ongoing controversial discussion about appropriate models, partitioning schemes, and assembly methods for phylogenomic alignments, coupled with the high computational cost to infer these, for many organismic groups, a sufficient number of taxa is often exclusively available from one or just a few genes (e.g., rbcL, matK, rDNA). In this paper we address scalability of Maximum-Likelihood-based phylogeny reconstruction with respect to the number of taxa by example of several large nested single-gene rbcL alignments comprising 400 up to 3,491 taxa. In order to test the effect of taxon sampling, we employ an appropriately adapted taxon jackknifing approach. In contrast to standard jackknifing, this taxon subsampling procedure is not conducted entirely at random, but based on drawing subsamples from empirical taxon-groups which can either be user-defined or determined by using taxonomic information from databases. Our results indicate that, despite an unfavorable number of sequences to number of base pairs ratio, i.e., many relatively short sequences, Maximum Likelihood tree searches and bootstrap analyses scale well on single-gene rbcL alignments with a dense taxon sampling up to several thousand sequences. Moreover, the newly implemented taxon subsampling procedure can be beneficial for inferring higher level relationships and interpreting bootstrap support from comprehensive analysis.

• Assessing the root of bilaterian animals with scalable phylogenomic methods.

  Authors: Andreas Hejnol, Matthias Obst, Alexandros Stamatakis, Michael Ott, Greg W Rouse, Gregory D Edgecombe, Pedro Martinez, Jaume Baguñà, Xavier Bailly, Ulf Jondelius, Matthias Wiens, Werner E G Müller, Elaine Seaver, Ward C Wheeler, Mark Q Martindale, Gonzalo Giribet, Casey W Dunn

  Proceedings. Biological sciences / The Royal Society. 09/2009;

  A clear picture of animal relationships is a prerequisite to understand how the morphological and ecological diversity of animals evolved over time. Among others, the placement of the acoelomorphA clear picture of animal relationships is a prerequisite to understand how the morphological and ecological diversity of animals evolved over time. Among others, the placement of the acoelomorph flatworms, Acoela and Nemertodermatida, has fundamental implications for the origin and evolution of various animal organ systems. Their position, however, has been inconsistent in phylogenetic studies using one or several genes. Furthermore, Acoela has been among the least stable taxa in recent animal phylogenomic analyses, which simultaneously examine many genes from many species, while Nemertodermatida has not been sampled in any phylogenomic study. New sequence data are presented here from organisms targeted for their instability or lack of representation in prior analyses, and are analysed in combination with other publicly available data. We also designed new automated explicit methods for identifying and selecting common genes across different species, and developed highly optimized supercomputing tools to reconstruct relationships from gene sequences. The results of the work corroborate several recently established findings about animal relationships and provide new support for the placement of other groups. These new data and methods strongly uphold previous suggestions that Acoelomorpha is sister clade to all other bilaterian animals, find diminishing evidence for the placement of the enigmatic Xenoturbella within Deuterostomia, and place Cycliophora with Entoprocta and Ectoprocta. The work highlights the implications that these arrangements have for metazoan evolution and permits a clearer picture of ancestral morphologies and life histories in the deep past.

• A reconfigurable architecture for the Phylogenetic Likelihood Function.

  Authors: Nikolaos Alachiotis, Alexandros Stamatakis, Euripides Sotiriades, Apostolos Dollas

  19th International Conference on Field Programmable Logic and Applications, FPL 2009, August 31 - September 2, 2009, Prague, Czech Republic; 01/2009

• Exploring FPGAs for accelerating the phylogenetic likelihood function.

  Authors: Nikolaos Alachiotis, Euripides Sotiriades, Apostolos Dollas, Alexandros Stamatakis

  23rd IEEE International Symposium on Parallel and Distributed Processing, IPDPS 2009, Rome, Italy, May 23-29, 2009; 01/2009

• Load Balance in the Phylogenetic Likelihood Kernel.

  Authors: Alexandros Stamatakis, Michael Ott

  ICPP 2009, International Conference on Parallel Processing, Vienna, Austria, 22-25 September 2009; 01/2009

• A rapid bootstrap algorithm for the RAxML Web servers.

  Authors: Alexandros Stamatakis, Paul Hoover, Jacques Rougemont

  Systematic biology. 11/2008; 57(5):758-71.

  Despite recent advances achieved by application of high-performance computing methods and novel algorithmic techniques to maximum likelihood (ML)-based inference programs, the major computationalDespite recent advances achieved by application of high-performance computing methods and novel algorithmic techniques to maximum likelihood (ML)-based inference programs, the major computational bottleneck still consists in the computation of bootstrap support values. Conducting a probably insufficient number of 100 bootstrap (BS) analyses with current ML programs on large datasets-either with respect to the number of taxa or base pairs-can easily require a month of run time. Therefore, we have developed, implemented, and thoroughly tested rapid bootstrap heuristics in RAxML (Randomized Axelerated Maximum Likelihood) that are more than an order of magnitude faster than current algorithms. These new heuristics can contribute to resolving the computational bottleneck and improve current methodology in phylogenetic analyses. Computational experiments to assess the performance and relative accuracy of these heuristics were conducted on 22 diverse DNA and AA (amino acid), single gene as well as multigene, real-world alignments containing 125 up to 7764 sequences. The standard BS (SBS) and rapid BS (RBS) values drawn on the best-scoring ML tree are highly correlated and show almost identical average support values. The weighted RF (Robinson-Foulds) distance between SBS- and RBS-based consensus trees was smaller than 6% in all cases (average 4%). More importantly, RBS inferences are between 8 and 20 times faster (average 14.73) than SBS analyses with RAxML and between 18 and 495 times faster than BS analyses with competing programs, such as PHYML or GARLI. Moreover, this performance improvement increases with alignment size. Finally, we have set up two freely accessible Web servers for this significantly improved version of RAxML that provide access to the 200-CPU cluster of the Vital-IT unit at the Swiss Institute of Bioinformatics and the 128-CPU cluster of the CIPRES project at the San Diego Supercomputer Center. These Web servers offer the possibility to conduct large-scale phylogenetic inferences to a large part of the community that does not have access to, or the expertise to use, high-performance computing resources.

• Efficient computation of the phylogenetic likelihood function on multi-gene alignments and multi-core architectures.

  Authors: Alexandros Stamatakis, Michael Ott

  Philosophical transactions of the Royal Society of London. Series B, Biological sciences. 10/2008;

  The continuous accumulation of sequence data, for example, due to novel wet-laboratory techniques such as pyrosequencing, coupled with the increasing popularity of multi-gene phylogenies and emergingThe continuous accumulation of sequence data, for example, due to novel wet-laboratory techniques such as pyrosequencing, coupled with the increasing popularity of multi-gene phylogenies and emerging multi-core processor architectures that face problems of cache congestion, poses new challenges with respect to the efficient computation of the phylogenetic maximum-likelihood (ML) function. Here, we propose two approaches that can significantly speed up likelihood computations that typically represent over 95 per cent of the computational effort conducted by current ML or Bayesian inference programs. Initially, we present a method and an appropriate data structure to efficiently compute the likelihood score on 'gappy' multi-gene alignments. By 'gappy' we denote sampling-induced gaps owing to missing sequences in individual genes (partitions), i.e. not real alignment gaps. A first proof-of-concept implementation in RAxML indicates that this approach can accelerate inferences on large and gappy alignments by approximately one order of magnitude. Moreover, we present insights and initial performance results on multi-core architectures obtained during the transition from an OpenMP-based to a Pthreads-based fine-grained parallelization of the ML function.

• Exploiting Fine-Grained Parallelism in the Phylogenetic Likelihood Function with MPI, Pthreads, and OpenMP: A Performance Study.

  Authors: Alexandros Stamatakis, Michael Ott

  Pattern Recognition in Bioinformatics, Third IAPR International Conference, PRIB 2008, Melbourne, Australia, October 15-17, 2008. Proceedings; 01/2008

• Multiple evolutionary mechanisms drive papillomavirus diversification.

  Authors: Marc Gottschling, Alexandros Stamatakis, Ingo Nindl, Eggert Stockfleth, Angel Alonso, Ignacio G Bravo

  Molecular biology and evolution. 05/2007; 24(5):1242-58.

  The circular, double-stranded 8-kb DNA genome of papillomaviruses (PVes) consists mainly of 4 large genes, E1, E2, L2, and L1. Approximately 150 papillomavirus genomes have been sequenced to date. WeThe circular, double-stranded 8-kb DNA genome of papillomaviruses (PVes) consists mainly of 4 large genes, E1, E2, L2, and L1. Approximately 150 papillomavirus genomes have been sequenced to date. We analyzed a representative sample of 53 PVes genomes using maximum likelihood, Bayesian inference, maximum parsimony, and distance-based methods both on nucleotide (nt) and on amino acid (aa) alignments. When the 4 genes were analyzed separately, aa-inferred phylogenies contradicted each other less than nt-inferred trees (judged by partition homogeneity tests). In particular, gene combinations including the L2 gene generated significant incongruence (P < 0.001). Combined analyses of the remaining genes E1-E2-L1 produced a well-supported phylogeny including supertaxon beta + gamma + pi + xi-PVes (infecting Artiodactyla, Carnivora, Primates, and Rodentia) and supertaxon kappa + lambda + mu + nu + sigma-PVes (infecting Carnivora, Lagomorpha, Primates, and Rodentia). Based on the tree topology, host-linked evolution appears plausible at shallow, rather than deeper, taxonomic levels. Diversification within PVes may also involve adaptive radiation establishing different niches (within a single-host species) and recombination events (within single-host cells). Heterogeneous groups of closely related PVes infecting, for example, humans and domestic animals such as hamster, dog, and cattle suggest multiple infections across species borders. Additional evolutionary phenomena such as strong codon usage preferences, and computational biases including reconstruction artifacts and insufficient taxon sampling, may contribute to the incomplete resolution of deep phylogenetic nodes. The molecular data globally supports a complex evolutionary scenario for PVes, which is driven by multiple mechanisms but not exclusively by coevolution with corresponding hosts.