Dataset Information

Improvement of domain-level ortholog clustering by optimizing domain-specific sum-of-pairs score.

ABSTRACT: BACKGROUND: Identification of ortholog groups is a crucial step in comparative analysis of multiple genomes. Although several computational methods have been developed to create ortholog groups, most of those methods do not evaluate orthology at the sub-gene level. In our method for domain-level ortholog clustering, DomClust, proteins are split into domains on the basis of alignment boundaries identified by all-against-all pairwise comparison, but it often fails to determine appropriate boundaries. RESULTS: We developed a method to improve domain-level ortholog classification using multiple alignment information. This method is based on a scoring scheme, the domain-specific sum-of-pairs (DSP) score, which evaluates ortholog clustering results at the domain level as the sum total of domain-level alignment scores. We developed a refinement pipeline to improve domain-level clustering, DomRefine, by optimizing the DSP score. We applied DomRefine to domain-level ortholog groups created by DomClust using a dataset obtained from the Microbial Genome Database for Comparative Analysis (MBGD), and evaluated the results using COG clusters and TIGRFAMs models as the reference data. Thus, we observed that the agreement between the resulting classification and the classifications in the reference databases is improved at almost every step in the refinement pipeline. Moreover, the refined classification showed better agreement than the classifications in the eggNOG databases when TIGRFAMs was used as the reference database. CONCLUSIONS: DomRefine is a useful tool for improving the quality of domain-level ortholog classification among microbial genomes. Combining with a rapid domain-level ortholog clustering method, such as DomClust, it can be used to create a high-quality ortholog database that can serve as a solid basis for various comparative genome analyses.

SUBMITTER: Chiba H

PROVIDER: S-EPMC4035852 | biostudies-other | 2014

REPOSITORIES: biostudies-other

ACCESS DATA

Publications

Improvement of domain-level ortholog clustering by optimizing domain-specific sum-of-pairs score.

Chiba Hirokazu H Uchiyama Ikuo I

BMC bioinformatics 20140518

<h4>Background</h4>Identification of ortholog groups is a crucial step in comparative analysis of multiple genomes. Although several computational methods have been developed to create ortholog groups, most of those methods do not evaluate orthology at the sub-gene level. In our method for domain-level ortholog clustering, DomClust, proteins are split into domains on the basis of alignment boundaries identified by all-against-all pairwise comparison, but it often fails to determine appropriate b ...[more]

PMID: 24885064

Similar Datasets

Project description:BackgroundThe need to expand and better engage patients in primary care improvement persists.PurposeRecognizing a continuum of forms of engagement, this study focused on identifying lessons for optimizing patient partnerships, wherein engagement is characterized by shared decision-making and practice improvement codesign.MethodologyTwenty-three semistructured interviews with providers and patients involved in improvement efforts in seven U.S. primary care practices in the Academic Innovations Collaborative (AIC). The AIC aimed to implement primary care improvement, emphasizing patient engagement in the process. Data were analyzed thematically.ResultsSites varied in their achievement of patient partnerships, encountering material, technical, and sociocultural obstacles. Time was a challenge for all sites, as was engaging a diversity of patients. Technical training on improvement processes and shared learning "on the job" were important. External, organizational, and individual-level resources helped overcome sociocultural challenges: The AIC drove provider buy-in, a team-based improvement approach helped shift relationships from providers and recipients toward teammates, and individual qualities and behaviors that flattened hierarchies and strengthened interpersonal relationships further enhanced "teamness." A key factor influencing progress toward transformative partnerships was a strong shared learning journey, characterized by frequent interactions, proximity to improvement decision-making, and learning together from the "lived experience" of practice improvement. Teams came to value not only patients' knowledge but also changes wrought by working collaboratively over time.ConclusionEstablishing practice improvement partnerships remains challenging, but partnering with patients on improvement journeys offers distinctive gains for high-quality patient-centered care.Practice implicationsEngaging diverse patient partners requires significant disruption to organizational norms and routines, and the trend toward team-based primary care offers a fertile context for patient partnerships. Material, technical, and sociocultural resources should be evaluated not only for whether they overcome specific challenges but also for how they enhance the shared learning journey.

Dataset Information

Improvement of domain-level ortholog clustering by optimizing domain-specific sum-of-pairs score.

Publications

Improvement of domain-level ortholog clustering by optimizing domain-specific sum-of-pairs score.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets