Table 2 Methods, tools, literature reviews and resources*

MANTRA transethnic meta-analysis software

Replication of primary signal in WA population and fine-mapping of second independent signal showing positive selection in WA, EA and EUR cohorts. MANTRA is available as a suite of executables on request from the author [58]. Major limitation in that it cannot estimate a joint effect size even for the combined meta-analysis

MANTRA [58]; applications: adiposity loci [59]; quantification of serum protein [14]; T2D [33]

RE-HE random-effects method

RE and FE models in the context of a meta-analysis with significant heterogeneity have low power. By relaxing overly conservative parameters in RE analysis algorithms, RE-HE provides more power in the presence of inter-study effect heterogeneity. Metasoft is available as a package [114]; it provides a joint effect size estimate, but it is the same as the RE estimate

RE-HE algorithm [56]; applications: endometriosis [115]; bipolar disorder [18]; multi-tissue eQTLs [116]

Review on replicability of transethnic association signals

Comprehensive review of literature across 28 diseases in EA and EUR populations demonstrating high replicability, sharing of disease alleles and good correlation of effect sizes

[43]

Review on power gains in meta-analytical approaches

Simulation-based analysis demonstrating that a multi-ethnic study design provides non-trivial power gains, especially when AFR populations are used to examine low frequency alleles (MAF <5%)

[117]

Comparative analysis of FE, RE, RE-HE and MANTRA as a method for GWAS meta-analysis

Results show that both RE-HE and MANTRA are computationally efficient and robust methods in accounting for effect size heterogeneity while providing a boost in power when compared with traditional meta-analysis methods. Results are provided for both simulations and application to T2D datasets

[57]

Modified RE-HE for joint analysis of resequencing data for rare variant gene-based analysis

Extension of RE-HE to provide a more powerful (than traditional RE) method to perform rare-variant burden testing in a heterogeneous resequencing study sample

[44]