Exogean: a framework for annotating protein-coding genes in eukaryotic genomic DNA. Export

@article{citeulike:878405, abstract = {BACKGROUND: Accurate and automatic gene identification in eukaryotic genomic DNA is more than ever of crucial importance to efficiently exploit the large volume of assembled genome sequences available to the community. Automatic methods have always been considered less reliable than human expertise. This is illustrated in the EGASP project, where reference annotations against which all automatic methods are measured are generated by human annotators and experimentally verified. We hypothesized that replicating the accuracy of human annotators in an automatic method could be achieved by formalizing the rules and decisions that they use, in a mathematical formalism. RESULTS: We have developed Exogean, a flexible framework based on directed acyclic colored multigraphs (DACMs) that can represent biological objects (for example, mRNA, ESTs, protein alignments, exons) and relationships between them. Graphs are analyzed to process the information according to rules that replicate those used by human annotators. Simple individual starting objects given as input to Exogean are thus combined and synthesized into complex objects such as protein coding transcripts. CONCLUSION: We show here, in the context of the EGASP project, that Exogean is currently the method that best reproduces protein coding gene annotations from human experts, in terms of identifying at least one exact coding sequence per gene. We discuss current limitations of the method and several avenues for improvement.}, address = {Dyogen Lab, CNRS UMR8541, Ecole Normale Sup\'{e}rieure, 46 rue d'Ulm, 75005 Paris, France.}, author = {Djebali, S. and Delaplace, F. and Crollius, H. R.}, citeulike-article-id = {878405}, citeulike-linkout-0 = {http://dx.doi.org/10.1186/gb-2006-7-s1-s7}, citeulike-linkout-1 = {http://view.ncbi.nlm.nih.gov/pubmed/16925841}, citeulike-linkout-2 = {http://www.hubmed.org/display.cgi?uids=16925841}, doi = {10.1186/gb-2006-7-s1-s7}, issn = {1465-6914}, journal = {Genome Biol}, keywords = {annotation, eukaryote, gene, method}, posted-at = {2006-09-30 00:21:18}, priority = {2}, title = {Exogean: a framework for annotating protein-coding genes in eukaryotic genomic DNA.}, url = {http://dx.doi.org/10.1186/gb-2006-7-s1-s7}, volume = {7 Suppl 1}, year = {2006} }

TY - JOUR ID - citeulike:878405 L3 - citeulike-article-id:878405 N2 - BACKGROUND: Accurate and automatic gene identification in eukaryotic genomic DNA is more than ever of crucial importance to efficiently exploit the large volume of assembled genome sequences available to the community. Automatic methods have always been considered less reliable than human expertise. This is illustrated in the EGASP project, where reference annotations against which all automatic methods are measured are generated by human annotators and experimentally verified. We hypothesized that replicating the accuracy of human annotators in an automatic method could be achieved by formalizing the rules and decisions that they use, in a mathematical formalism. RESULTS: We have developed Exogean, a flexible framework based on directed acyclic colored multigraphs (DACMs) that can represent biological objects (for example, mRNA, ESTs, protein alignments, exons) and relationships between them. Graphs are analyzed to process the information according to rules that replicate those used by human annotators. Simple individual starting objects given as input to Exogean are thus combined and synthesized into complex objects such as protein coding transcripts. CONCLUSION: We show here, in the context of the EGASP project, that Exogean is currently the method that best reproduces protein coding gene annotations from human experts, in terms of identifying at least one exact coding sequence per gene. We discuss current limitations of the method and several avenues for improvement. JF - Genome Biol SN - 1465-6914 AD - Dyogen Lab, CNRS UMR8541, Ecole Normale Supérieure, 46 rue d'Ulm, 75005 Paris, France. TI - Exogean: a framework for annotating protein-coding genes in eukaryotic genomic DNA. VL - 7 Suppl 1 KW - annotation KW - eukaryote KW - gene KW - method AU - Djebali, S AU - Delaplace, F AU - Crollius, HR PY - 2006/// UR - http://dx.doi.org/10.1186/gb-2006-7-s1-s7 ER -