Validate.Rmd

---
title: "External Validate"
---

# Example data
- DBSLMM result (`your/path/out/summary_gemma_chr1.assoc.dbslmm.txt`) <br>
The result of DBSLMM.  <br>
- External summary statistics (`summary_external.txt`) <br>
The summary statsitics of standing hright (GIANT data).
- Reference data (`ref_chr1.bed`, `ref_chr1.bim` and `ref_chr1.fam`) <br>
We sampled 300 samples from the 1000 Gemone Project data.  <br>
Note: The phenotype of test data is simulated by normal distribution. Therefore, you can not use the test data to evaluate the prediction performance.

# Input file format
- DBSLMM result <br>
````{r, engine = 'bash', eval = FALSE}
rs13302957 G 0.546652 1.82642 1
rs3748588 T 0.273432 1.64563 1
rs74045047 A -0.282306 -0.5378 1
rs3845292 G 0.159231 0.22554 1
rs113288277 T -0.0524222 -0.18273 1
rs112797925 A 0.252739 1.12303 1
rs12743678 A 0.741517 1.51564 1
rs141242758 C 0.00548007 0.0124342 0
rs28544273 A 0.0114189 0.0248473 0
rs3115860 C 0.00628248 0.013297 0
````
- External summary statistics <br>
````{r, engine = 'bash', eval = FALSE}
rs4747841 A 0.449 -0.000773760673593586
rs4749917 T 0.436 0.000771419263435909
rs878177 T 0.3 0.009073036977771
rs12219605 T 0.427 0.00076948282631908
rs3763688 C 0.144 -0.00109233489370248
rs3763689 T 0.217 -0.00466354028609167
rs1983867 C 0.431 0.00770375479879779
rs1983866 A 0.288 -0.0089655982510929
rs1983865 T 0.425 0.00769017229976026
rs1983864 T 0.3 -0.00907303697777101
````
- Reference panel ([PLINK format](https://www.cog-genomics.org/plink/1.9/formats)) <br>
The same file name of bed, bim and fam files. 

# Input parameters
We prepare all the input file in the folder `test_dat`. You can output the explanation for each parameter: 
````{r, engine = 'bash', eval = FALSE}
Rscript VALID.R -h
Rscript VALID.R --help
````
The details is: 
````{r, engine = 'bash', eval = FALSE}
--dbslmm=CHARACTER
		INPUT:  the result of dbslmm
--external=CHARACTER
		INPUT:  the external summary statistics
--ref=CHARACTER
		INPUT:  the perfix of reference panel
--block=CHARACTER
		INPUT:  the block information (Berisa and Pickrell 2015)
--chr=CHARACTER
		INPUT:  the chromosome number
--mafMax=CHARACTER
		INPUT:  the maximium of the difference between reference panel and external data
--outPath=CHARACTER
		OUTPUT: the output path
````

# Output
The output of `VALID.R` is the $R^2$. It will be print in the screen. 

# Example code
We use the data of `test_dat` to show the external validation. 
````{r, engine = 'bash', eval = FALSE}
### change file permission 
valid=/your/path/DBSLMM/software/valid
chmod 777 ${valid}
### External validation
VALID=/your/path/DBSLMM/software/VALID.R
dbslmmRes=/your/path/DBSLMM/test_dat/out/summary_gemma_chr
summf=/your/path/DBSLMM/test_dat/external_summary
outPath=/your/path/DBSLMM/test_dat
ref=/your/path/DBSLMM/test_dat/ref_chr
blockf=/your/path/DBSLMM/test_dat/chr
### for chromosome 1 (example)
Rscript ${VALID} --dbslmm ${dbslmmRes} --external ${summf}.txt --ref ${ref} --valid ${valid} --block ${blockf} \
                 --chr 1 --mafMax 1 --outPath ${outPath}
### for whole genome (If you have whole genome data, you can use the command)
Rscript ${VALID} --dbslmm ${dbslmmRes} --external ${summf}.txt --ref ${ref} --valid ${valid} --block ${blockf} \
                 --chr all --mafMax 0.2 --outPath ${outPath}
````