The Lattice Project

From BOINC Projects
Revision as of 12:22, 14 July 2026 by Al Piskun (talk | contribs) (first light)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search



The Lattice Project
Project
StatusCompleted
CategoryBioinformatics, Computational biology, Phylogenetics
ComputeCPU
Development
DeveloperAdam L. Bazinet
AuthorMichael P. Cummings (principal investigator); Adam L. Bazinet (lead developer)
SponsorUniversity of Maryland Institute for Advanced Computer Studies (UMIACS); Center for Bioinformatics and Computational Biology, University of Maryland
MaintainerNone (discontinued)
Initial releaseDecember 18, 2007  (19 years ago)
CompletedMay 17, 2017  (9 years ago) (exact date unrecorded)
Software
Written inJava (Grid middleware); C++ (scientific applications such as GARLI)
Operating systemWindows, Linux, macOS
BOINC statistics
Stats as ofJune 16, 2014  (12 years ago)
Total users17,000+
Total hosts38,740
Metadata
Websitehttp://lattice.umiacs.umd.edu/ and 2nd URL - http://boinc.umiacs.umd.edu/
LicenseNot uniformly licensed; some constituent applications (e.g. GARLI) distributed as open source
The Lattice Project was developed and administered at the University of Maryland, College Park.

The Lattice Project was a discontinued grid computing and volunteer computing system operated by the Laboratory of Molecular Evolution at the University of Maryland, College Park.[1] Rather than being a single program, Lattice combined several distinct computing resources, including dedicated Condor pools, PBS compute clusters, and a public BOINC volunteer pool, into one production Grid system built on the Globus Toolkit.[2] Its best-known use was distributing maximum-likelihood phylogenetic analyses through the Genetic Algorithm for Rapid Likelihood Inference (GARLI) program, letting researchers worldwide reconstruct evolutionary relationships from DNA and protein sequence data.[1] Because the project offered two distinct, separately administered instances, "the Lattice Grid" and "the Lattice BOINC Project", it is described here in both of its forms below.

History

The system that became the Lattice Project grew out of earlier commodity Grid computing experiments at Maryland's Center for Bioinformatics and Computational Biology, first described by David Myers and Michael Cummings in a 2003 paper on combining Condor pools with simple job-submission tools.[3] That work developed into a full production Grid, described publicly by Cummings in 2005 and formally written up by Bazinet and Cummings in a 2008 book chapter on combining multiple Grid computing models.[4]

The volunteer-facing BOINC instance was formally opened to the public in a BOINC News announcement by project developer David Anderson, who described Lattice as supplying computing power for research into evolutionary relationships from DNA sequence data, bacterial and viral protein sequences, and biological diversity in nature reserves.[5] Bazinet's 2009 master's thesis subsequently documented the combined architecture in detail, describing a base of four Condor pools, three PBS clusters, and one BOINC pool at that time.[2]

In 2010, researchers at Maryland and the University of Arizona layered a public web gateway, molecularevolution.org, on top of the Grid, giving anyone with a browser access to Lattice's phylogenetic computing power rather than requiring the BOINC client.[6] By 2014, the university reported that more than 17,000 volunteers from 146 countries had contributed computing time to the project, with 38,740 personal computers registered to the BOINC pool alongside institutional resources at UMIACS, Coppin State University, and the Smithsonian Institution.[7] That same year Bazinet, Zwickl, and Cummings formally described the GARLI 2.0 web service built atop the Grid,[8] and in 2015 Bazinet and Cummings published a follow-up study on subdividing long BOINC workunits to improve throughput on the Lattice BOINC Project specifically.[9] A March 2015 forum thread on BOINCstats/BAM! confirms that the project's servers were still reporting recent throughput at that time.[10]

The project was eventually discontinued. According to the English Wikipedia's account, the Lattice BOINC website went offline, and because BOINC clients depend on a project's website to retrieve setup and scheduling information, the loss of that site effectively shut the volunteer computing instance down.[1] The exact discontinuation date was not documented in any source found for this article, but the site was already inactive by the time the Wikipedia article's supporting citation was archived in September 2017.[1]

The Lattice Grid

The Lattice Grid combined dedicated Condor pools with PBS clusters and the BOINC volunteer pool under a single Globus-based scheduling layer.

The Lattice Grid, reachable at lattice.umiacs.umd.edu, was the research and production Grid environment underlying the whole project. It ran the full Globus Toolkit software stack, including a web services container for deploying Grid-enabled applications, and used the Grid Security Infrastructure for authentication and authorization across participating resources.[2] At the time of Bazinet's 2009 thesis, the Grid drew on roughly 4,000 to 5,000 CPUs contributed by groups within the University of Maryland (UMIACS, the Office of Information Technology, and other campus units) as well as external partners such as Coppin State University.[2]

Rather than replacing existing campus computing infrastructure, the Lattice Grid was designed to federate it: local resources running Condor or PBS could be tied into the Grid by installing Globus on a submission node, after which jobs could be routed to whichever available resource, dedicated cluster, Condor pool, or the BOINC volunteer pool, suited the workload.[2] The Grid server itself handled transitory job storage and coordinated data transfer over GridFTP, while a separate "sister" server, contemporaneously described as running with 8 GB of RAM, was dedicated specifically to powering the BOINC-facing project.[2]

The Lattice BOINC Project

The Lattice BOINC Project, at boinc.umiacs.umd.edu, was the public-facing volunteer computing instance of the Grid and the piece most familiar to ordinary BOINC users. It was announced to the wider BOINC community in December 2007 as being open to volunteers, supplying computing power to researchers studying evolutionary relationships from DNA sequence data, protein sequence comparison, and even conservation reserve network design.[5] Community-maintained BOINC project listings describe it as running on Macintosh, Linux, and Windows, without GPU acceleration and without an OpenGL screensaver, distinguishing it from many contemporaneous BOINC projects that had already added GPU support.[11] The project was also listed among the participating projects on the volunteer computing aggregator GridRepublic, which summarized it simply as supporting the work of researchers at the University of Maryland.[12]

Unlike many single-application BOINC projects, the Lattice BOINC Project was designed from the outset to serve multiple scientific applications rather than one; David Anderson's original announcement noted that volunteers might receive workunits from a variety of different research applications rather than a single program, a design choice acknowledged in the project's own documentation as something of a public-relations challenge, since users could not always predict what kind of task their computer would be asked to run.[2]

Software and applications

The flagship application run on the Lattice Grid was GARLI, the Genetic Algorithm for Rapid Likelihood Inference, an open-source phylogenetic inference program written and maintained by Derrick J. Zwickl that searches for the evolutionary tree, branch lengths, and substitution model parameters maximizing the likelihood of a sequence dataset.[8][13] GARLI's underlying approach was first developed in Zwickl's 2006 doctoral dissertation at the University of Texas at Austin.[14] In general terms, a maximum-likelihood phylogenetic search of this kind seeks the tree topology T, branch lengths 𝐯, and model parameters θ that maximize

L(T,𝐯,θ∣X)=∏i=1nP(Xi∣T,𝐯,θ)

over the columns Xi of a sequence alignment X, a computation that becomes extremely demanding as the number of taxa grows, which is precisely the kind of workload the Lattice Grid was built to distribute across thousands of processors.[8]

An example of a maximum-likelihood phylogenetic tree of the general kind produced by GARLI analyses distributed across the Lattice Grid.

Beyond GARLI, Bazinet's 2009 thesis documents other Grid-enabled applications run on Lattice, including HMMPfam for protein sequence comparison and MARXAN for conservation reserve network design, alongside a number of older research applications from the project's earlier years.[2] By the project's own account, roughly 25 applications had been Grid-enabled over its lifetime, though only a subset saw significant production use.[2]

Research output

Analyses run through the Lattice Grid and its GARLI web service contributed to a range of published phylogenetic studies, particularly in the University of Maryland's own moth and butterfly research program. Examples include a 2009 study on the phylogeny of advanced moths and butterflies (Lepidoptera: Ditrysia), a 2011 study on gene sampling in gracillariid leaf-mining moths, and a 2013 exploratory RNA-Seq study of rapid radiation within Apoditrysia, all of which drew on GARLI analyses of the kind distributed by Lattice.[8]

Discontinuation

The Lattice Project is not currently listed among active BOINC projects, and both of its web addresses, lattice.umiacs.umd.edu and boinc.umiacs.umd.edu, are offline as of this writing. Because BOINC client software depends on a project's own website to retrieve account, application, and scheduling information, the loss of the Lattice BOINC website meant volunteer computers could no longer attach to or receive work from the project, effectively ending that instance regardless of whether any Grid infrastructure remained active elsewhere.[1]

See also

References

  1. ↑ 1.0 1.1 1.2 1.3 1.4 The Lattice Project. Wikipedia. Retrieved 2026-07-14.
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 Template:Cite thesis
  3. ↑ (2003).Necessity is the mother of invention: a simple grid computing system using commodity tools. Journal of Parallel and Distributed Computing. pp. 578-589.
  4. ↑ "The Lattice Project: a Grid research and production environment combining multiple Grid computing models". In:(2008).Distributed & Grid Computing: Science Made Transparent for Everyone. Principles, Applications and Supporting Communities. Rechenkraft.net. pp. 2-13.Edited by Weber, Marc H. W..Marburg.
  5. ↑ 5.0 5.1 Anderson, David.(2007-12-18).The Lattice Project, based at the University of Maryland Center for Bioinformatics and Computational Biology, is now open to volunteers. BOINC. Retrieved 2026-07-14.
  6. ↑ UMD-developed Infrastructure Offers Hardware, Software for Phylogenetic Analysis. GenomeWeb. Retrieved 2026-07-14.
  7. ↑ Cummings, Michael.(2014-06-16).UMD-led Public Computational Biology Web Service Gains Popularity. University of Maryland College of Computer, Mathematical, and Natural Sciences. Retrieved 2026-07-14.
  8. ↑ 8.0 8.1 8.2 8.3 (2014).A Gateway for Phylogenetic Analysis Powered by Grid Computing Featuring GARLI 2.0. Systematic Biology. pp. 812-818. DOI: 10.1093/sysbio/syu031.
  9. ↑ (2015).Subdividing Long-Running, Variable-Length Analyses Into Short, Fixed-Length BOINC Workunits. Journal of Grid Computing. DOI: 10.1007/s10723-015-9348-5.
  10. ↑ Forum::New projects::Lattice. BOINCstats/BAM!. Retrieved 2026-07-14.
  11. ↑ The Lattice Project. BOINC Wiki (Mundayweb). Retrieved 2026-07-14.
  12. ↑ GridRepublic: About. GridRepublic. Retrieved 2026-07-14.
  13. ↑ Home. GARLI wiki, GitHub. Retrieved 2026-07-14.
  14. ↑ Template:Cite thesis
Retrieved from "https://boincsynergy.ca/wiki/index.php?title=The_Lattice_Project&oldid=1710"