Unlocking the potential of survival data for model organisms through a new database and online analysis platform: SurvCurv.
Bottom Line: Understanding the biology of aging is highly desirable because of the benefits for the wide range of aging-related diseases.The database, available at www.ebi.ac.uk/thornton-srv/databases/SurvCurv/, offers various functions including plotting, Cox proportional hazards analysis, mathematical mortality models and statistical tests.It facilitates reanalysis and allows users to analyse their own data and compare it with the largest repository of model-organism data from published experiments, thus unlocking the potential of survival data and demographics in model organisms.
Affiliation: EMBL - European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SD, UK.Show MeSH
Related in: MedlinePlus
Mentions: SurvCurv also offers the possibility to analyse database data or uploaded data using the Cox proportional hazards (CoxPH) model (Cox, 1972), a statistical model of survival data with one or more covariates or factors, that is, for multiple conditions. It allows the user to identify which factors significantly contribute to the overall model and quantify their respective influences on the hazard rates in the model. An increased survival corresponding to a decreased mortality or hazard rate is indicated by an exp(coef) < 1, with exp(coef) giving the relative mortality risk compared to the baseline. Importantly, the CoxPH model assumes that the hazard rates of the different conditions are proportional, that is, the mortality of all conditions are multiples of each other. This assumption thus has to be checked before each CoxPH analysis. SurvCurv automatically tests the assumption and presents the results of the tests, together with the CoxPH results and diagnostic plots of the assumption (see Fig. 4 for an example). It has to be noted that p-values of the proportional hazards tests, like of any other test, are strongly dependent on the sample size. Gross violations may not be statistically significant if the sample size is very small, and even slight violations, causing negligible errors in the estimated coefficients, may be highly significant if the sample size is very large. An estimate of the size of the deviations from the assumed independence, that is, no correlation (see Methods for details), is given by the respective correlation coefficients rho, which can thus be helpful in interpreting the test results. If the proportional hazards assumption has been rejected with noticeable deviations, alternative analysis such as accelerated failure time analysis (AFT) or Cox analysis using transformed or time-dependent co-variates might be necessary. These are currently not available in SurvCurv, but users requiring these analyses can download all data for use in their preferred full statistics program.
Affiliation: EMBL - European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SD, UK.