Identifying phenotypic correlations between complex traits and diseases can provide useful etiological insights and help correct multiple testing. Lack of centralized individual-level phenotypes database makes it almost impossible to estimate phenotypic correlations across human traits and diseases as a whole picture. A useful alternative is to use the genome-wide association study (GWAS) summary statistics to estimate phenotypic correlations via the method metaCCA. We applied this method to the centralized GWAS summary results database we created to estimate 358,281 phenotypic correlations among 846 traits and diseases. The atlas of phenotypic correlations systematically scan hypotheses across a large scale of human traits and diseases, which can further tested using methods such as Mendelian randomization, LD score regression and PheWAS. The matrix also informed the selection of covariates for genetic, epigenetic and epidemiology analysis. In addition, we compared the phenotypic correlation and genetic correlation amongst 173 traits. The results of metaSpD suggest a 562 number of independent variable across 846 traits and diseases (P-value threshold of 9e-05). Additionally, metaSpD includes principal-component analysis which enables selection of a subset of traits in a complex molecular network, e.g. metabolites.