LiteQTL.jl Documentation
Package information
LiteQTL is a package that runs whole genome QTL scans near real-time, utilizing the computation power of GPU.
Features
- Near real time computation for whole genome scan using Linear Model
- Genome scan with covairates
- CPU parallelization and GPU acceleration
- Input data can be of different precisions (Float32, or Float64)
Input and Output
Input (all with no missing data)
- Genotype probability
- Phenotype
- Covariates (Optional)
Output
- (Default) Maximum LOD (Log of Odds) score, and the index of the maximum
- LOD (Log of Odds) matrix
Example
We have created a Jupyter notebook with the whole analysis pipeline using the BXD spleen data as an example. Users can run eQTL scans and create a eQTL hits figure. See: example/spleen_analysis.ipynb
Auxilary Github Repositories
This repo contais scripts to compile the LiteQTL package to remove the compilation time of Julia (the extra time in the first run in Julia REPL).
It is an effort to make our research reproducible. All code related to experiment reuslt, from dowloading data, cleaning data, to running LiteQTL and creating figure are found in this repository. You can recreate the results in our paper Speeding up eQTL scans in the BXD population using GPUs using the scripts in this repository.
Index
LiteQTL.calculate_nr
LiteQTL.calculate_nr
LiteQTL.calculate_px
LiteQTL.cpurun
LiteQTL.filter_maf
LiteQTL.find_max_idx_value
LiteQTL.get_geno_data
LiteQTL.get_pheno_block_size
LiteQTL.get_pheno_data
LiteQTL.get_standardized_matrix
LiteQTL.get_standardized_matrix_gpu
LiteQTL.gpu_square_lod
LiteQTL.gpurun
LiteQTL.is_corr_in_range
LiteQTL.lod2p
LiteQTL.lod_kernel
LiteQTL.lod_score_multithread
LiteQTL.lod_score_multithread
LiteQTL.reduce_kernel
LiteQTL.scan
LiteQTL.set_blas_threads