MultiSeq: unifying sequence and structure data for evolutionary analysis.

Roberts E, Eargle J, Wright D, Luthey-Schulten Z. · 2006

Read MultiSeq: unifying sequence and structure data for evolutionary analysis. by Roberts E, Eargle J, Wright D, Luthey-Schulten Z. (2006) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc1586216.txt.

Introduction

Evolutionary studies drown in sequence and structure data held apart. Roberts, Eargle, Wright, and Luthey-Schulten built MultiSeq in 2006 to unify them. Public databases then held over 3 million sequences and 35,000 structures.

MultiSeq organizes, aligns, and displays both data kinds for proteins and nucleic acids in one space. It carries standard classifications that users can adjust in an electronic notebook. A new algorithm builds clean evolutionary profiles for families of distant kin, stripping redundancy from alignments. Roberts frames the tool around evolution, but built it flexible enough for other patterns of use.

Catalog record

CREATORRoberts E, Eargle J, Wright D, Luthey-Schulten Z.
YEAR2006
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC1586216

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

MultiSeq: unifying sequence and structure data for evolutionary analysis By Elijah Roberts, John Eargle, Dan Wright, Zaida Luthey-Schulten Published in: BMC Bioinformatics DOI: 10.1186/1471-2105-7-382 ABSTRACT Abstract Background Since the publication of the first draft of the human genome in 2000, bioinformatic data have been accumulating at an overwhelming pace. Currently, more than 3 million sequences and 35 thousand structures of proteins and nucleic acids are available in public databases. Finding correlations in and between these data to answer critical research questions is extremely challenging. This problem needs to be approached from several directions: information science to organize and search the data; information visualization to assist in recognizing correlations; mathematics to formulate statistical inferences; and biology to analyze chemical and physical properties in terms of sequence and structure changes. Results Here we present MultiSeq, a unified bioinformatics analysis environment that allows one to organize, display, align and analyze both sequence and structure data for proteins and nucleic acids. While special emphasis is placed on analyzing the data within the framework of evolutionary biology, the environment is also flexible enough to accommodate other usage patterns. The evolutionary approach is supported by the use of predefined metadata, adherence to standard ontological mappings, and the ability for the user to adjust these classifications using an electronic notebook. MultiSeq contains a new algorithm to generate complete evolutionary profiles that represent the topology of the molecular phylogenetic tree of a homologous group of distantly related proteins. The method, based on the multidimensional QR factorization of multiple sequence and structure alignments, removes redundancy from the alignments and orders the protein sequences by increasing linear dependence, resulting in the identification of a minimal basis set of sequences that spans the evolutionary space of the homologous group of proteins. Conclusion MultiSeq is a major extension of the Multiple Alignment tool that is provided as part of VMD, a structural visualization program for analyzing molecular dynamics simulations. Both are freely distributed by the NIH Resource for Macromolecular Modeling and Bioinformatics and MultiSeq is included with VMD starting with version 1.8.5. The MultiSeq website has details on how to download and use the software: http://www.scs.uiuc.edu/~schulten/multiseq/ BACKGROUND In the field of bioinformatics, research activities are often split into two distinct areas: sequence analysis or structure analysis. Genomic and other sequencing projects generate enormous amounts of sequence data that are initially released with large portions annotated as either putative or hypothetical. Structural data, even in the era of structural genomics, are produced at a slower pace but analyzed to a high degree before being deposited into the public databases, such as PDB [1], SCOP [2,3]/Astral [4], CATH [5] and NDB [6]. This difference in pace has led to an increasing discrepancy in the relative sizes of these two data sets. The total size of the sequence databases (NCBI [7,8], EMBL [9], DDBJ [10], JGI [11], RDP [12], Swiss-Prot/TrEMBL [13], CRW [14], Bayreuth tRNA compilations [15], and the Genomic tRNA Database [16]) is several orders of magnitude greater than that of the structure databases. For a given protein, the large number of available sequences allows more complete evolutionary analyses. Multiple sequence alignments (MSA) are instrumental in identifying key conserved areas of a sequence, developing an evolutionary history of a molecule [17], and examining the covariance within a sequence in response to evolutionary pressure [18]. These analyses depend on having enough sequences to perform a well-balanced statistical analysis. The advantages of structural data are that they provide much more detailed information about the molecule in question, allowing specific atomic level interactions to be analyzed. Additionally, since structure is more conserved than sequence [19], structural data can be used to reconstruct many of the deeper evolutionary branches that would be difficult or impossible to determine with sequence data alone [20-23]. VMD [24], which currently has more than 30,000 registered users, provides powerful visualization and analysis capabilities for both structural and dynamics data generated from molecular dynamics simulations, as well as energetics derived from molecular mechanics force fields. It is optimized to handle large scale systems containing millions of atoms. VMD also implements a flexible scripting interface for the creation of custom tools. The previous Multiple Alignment [25] extension to VMD added only the capability to use evolutionary information obtained from multiple structures for interpreting structural results. Our goal with MultiSeq is to extend VMD's capabilities further by incorporating the more diverse evolutionary data available in sequences into the analysis process. There are already a large number of tools available to analyze bioinformatic data, but, like the field itself, they are mostly segregated into either sequence or structure tools. In the sequence world there are tools for viewing, analyzing, and editing an MSA like AE2 [26], CINEMA [27], ClustalX [28], and Jalview [29]; there are tools for creating MSAs by aligning individual sequences and profiles, like ClustalW [30], HMMER [31] and T-Coffee [32]; there are tools for annotating sequence data such as Pfaat [33]; and BLAST [34] is used for searching through databases for related sequences. One popular package, MEGA3 [35], provides an evolutionary approach to analyzing protein and nucleic acid sequences, including many easy to use features for determining sequence based phylogenies. Similarly, in the structure world there are numerous tools for visualizing structural data and performing structural analyses, including RASMOL [36], STAMP [37], STRIDE [38], and 3DNA [39]. There are also a few programs that combine sequence and structure data either for specific purposes, such as Swiss-PdbViewer/SWISS-MODEL [40] and MolIDE/SCWRL3 [41,42] for homology modeling, or as part of a pre-computed database of attributes, of which STING [43] is the primary example. Modeller [44] allows structural features to be built using sequences and structures of homologous proteins and nucleic acids. UCSF Chimera [45], a well known molecular modeling program that originated to handle small molecule docking, provides the ability to use sequence data in conjunction with structural data. However, it lacks some of the features needed to perform well-balanced evolutionary analyses, such as phylogenetic tree construction and elimination of bias. Friend [46], a bioinformatics application, has many of the sequence features required for performing evolutionary analyses, but has insufficient structural functionality to fully interpret the results in a structural context. InsightII and Discovery Studio (Accelrys Inc.) and MOE (Chemical Computing Group Inc.) are popular commercial packages for analyzing protein/drug interactions based on protein structure, dynamics, and energetics. Both programs can also use sequence data to perform combined analyses. NCBI's Cn3D [47] also supports both sequence and structure data, although it is primarily designed for use with pre-computed 3D superpositions and MSAs. The complementary information provided by fusion of these two data sources can give insight into evolutionary changes in sequence, structure, and function. However, the conceptual spaces of these fields differ, often resulting in mutual incomprehensibility to researchers in each field. MultiSeq, in dealing with both sequence and structure data in a way that is accessible to both areas, helps to bridge this gap. It does so through the use of integrated cross-referencing that acts as an informal version of ontology-driven knowledge extraction and discovery [48]. We plan to enhance future versions of MultiSeq to incorporate formal ontological methods, including using the work of groups such as the Gene Ontology project [49]. There is tremendous utility

Read the complete text