shu-hai / fdrhmrf

A FDR controlling procedure based on hidden Markov random field (Biometrics-15 paper)

FDRhmrf

This package is designed for the false discovery rate (FDR) controlling procedure proposed in [1]. This FDR procedure is based on the two-parameter Ising model (a classical hidden Markov random field) and the local index of significance (LIS; [2,3]), which aims to minimize the false nondiscovery rate (FNR) while controlling FDR at a given level especially for 3D spatial data, e.g., neuroimaging data.

This package is coded in C++, and can be called from Matlab by MEX. Examples are provided, and please see README.pdf for details.

Please cite the article [1] for this package, which is available here.

[1] Shu, H., Nan, B., and Koeppe, R. (2015). "Multiple Testing for Neuroimaging via Hidden Markov Random Field". Biometrics, 71, 741-750.

@article{shu2015multiple,
  title={Multiple testing for neuroimaging via hidden Markov random field},
  author={Shu, Hai and Nan, Bin and Koeppe, Robert},
  journal={Biometrics},
  volume={71},
  number={3},
  pages={741--750},
  year={2015},
  publisher={Wiley Online Library}
}

[2] Sun, W., and Cai. T.T. (2009). "Large‐scale multiple testing under dependence". Journal of the Royal Statistical Society, Series B, 71, 393-424.

[3] Wei, Z., Sun, W., Wang, K., and Hakonarson, H. (2009). "Multiple testing in genome-wide association studies via hidden Markov models". Bioinformatics 25, 2802–2808.