Introduction of retrospective saddlepoint approximation (retrospective SPA) approach in GWAS

What is retrospective SPA?

Retrospective saddlepoint approximation (Retrospective SPA) is a method applied in GWAS. For a score statistic (S=GTR), retrospective SPA strategy

  • (1) considers genotypes as random variables.
  • (2) uses saddlepoint approximation (SPA) to accurately approximate the null distribution of score statistics conditional on phenotype and covariates.

Main features of GWAS methods based on retrospective SPA strategy

  • Applicable to a wide range of trait types (binary, quantitative, time-to-event, ordinal, longitudinal, and other traits).

  • Maintain high accuracy when testing low-frequency or rare variants, even when the phenotypic distribution is unbalanced (e.g. case-control imbalance in case-control studies).

Citation of retrospective SPA

Retrospective saddlepoint approximation (Retrospective SPA) methods were first proposed in the master’s thesis:

Based on the idea from the authors of the above master’s thesis, we have applied retrospective saddlepoint approximation to several methods including:

If you utilized the retrospective saddlepoint approximation method in your proposed methods or tools, please acknowledge and respect the original ideas presented in the two works (SPAmix and SPAGxE). Additionally, kindly cite the original papers (SPAmix and SPAGxE) or the master’s thesis:

in accordance with academic standards.

Questions about retrospective SPA

Given that the two pivotal papers

  • (1) SPAmix: A scalable, accurate, and universal analysis framework using individual-level allele frequency for large-scale genetic association studies in an admixed population

  • (2) SPAGxE: A scalable and accurate framework for large-scale genome-wide gene-environment interaction analysis and its application to time-to-event and ordinal categorical traits

have not yet been published–despite the retrospective saddlepoint approximation method being proposed in 2021 (first proposed in 2021 and published in the master thesis of 马雨茁.经验鞍点近似方法及其在全基因组关联分析中的应用研究.2022.山东大学,MA thesis.doi:10.27272/d.cnki.gshdu.2022.002946., please consult the relevant authors and supervisors the reasons for the delay in publication.

Suggestions or comments on retrospective saddlepoint approximation methods are also welcome.

Timeline of the Retrospective SPA Project

March 2021

  • Event: The idea of a retrospective saddlepoint approximation (SPA) was conceived.

October 2021

  • Event: The SPAmix algorithm was successfully designed.

October 2021 - February 2022

  • Event: Work on the SPAGxE algorithm was initiated.

May 2022

  • Event: The master’s thesis (DOI:10.27272/d.cnki.gshdu.2022.002946), which initially introduced the concept of retrospective SPA, was completed.

RetroSPA_timeline