R package supporting the paper "ENGEP: advancing spatial transcriptomics with accurate unmeasured gene expression prediction".
ENGEP combins multiple prediction results from different reference datasets and prediction methods using a weighted average ensemble strategy to predict expression levels of spatially unmeasured genes. ENGEP mainly includes two steps: (i) generating multiple base results using k-nearest-neighbor (k-NN) regression. Different reference datasets, similarity measures, and numbers of neighbors (k) are used for this step. (ii) Combining these base results into a consensus result using a weighted average ensemble strategy. In this step, weights are assigned to different reference datasets to take into account their predictive power.
The source code is also deposited in Zenodo with a DOI assignment (DOI: https://doi.org/10.5281/zenodo.8365572).
install.packages("devtools")
devtools::install_github("Zhangxf-ccnu/ENGEP")
Note that package ‘propr’ was removed from the CRAN repository, we advise that the user should install propr before install ENGEP.
devtools::install_github("tpq/propr")
A tutorial with examples of the usage of ENGEP
is available at: ENGEP-examples.html.
Please do not hesitate to contact Miss Yang Shi-Tong styang@mails.ccnu.edu.cn or Dr. Xiao-Fei Zhang zhangxf@mail.ccnu.edu.cn to seek any clarifications regarding any contents or operation of the archive.