Releases: choishingwan/PRSice
Releases · choishingwan/PRSice
New Year Update: First Major Update of PRSice
Description
This is the first ever major update of PRSice 2 where there are a number of changes made where some of the old command line options are changed (e.g. --cov-header
is now depreciated and replaced by --cov-col
for consistency)
Below are the detail log of the changes made
Update Info
- Clumping algorithm has been completely changed. We should no longer encounter memory problem without receiving a meaningful error message. We have also replace most of our clumping code with the PLINK algorithm
- Original
--score
option is now renamed to--missing
to better resemble it's core function (as in how to interpret missing score) --score
option is now used to determine which kind of PRS to be calculated, (averaging (default), sum only or standardise)- PRSice now support gz file as base
- Fix some problems related to the reference LD panel
--all
is now named as--all-score
for better clarity- We changed the default behaviour of PRSice to always include the p-value threshold of 1. Therefore we replace
--full
with--no-full
to account for this change. - Introducing the
--no-default
flag. This will forbid PRSice from guessing the data field from the base file. So for example, the BP column will not be included if user use the--no-default
flag and didn't provide the--bp BP
option - Introducing the
--maf
,--geno
,--info
,--ld-maf
,--ld-geno
and--ld-info
flag which allow direct genotyping QC to be performed on the target data (NOTE: We have only test run these flags on the binary plink format. So it would be nice if user help us to test run these for the bgen file) - You can now filter SNPs from the base file using the
--info-base
and--maf-base
option - PRSice now comes with a window version. To enable that, a number of changes have been made:
- pthread instead of std::thread is used.
- Replace p-value calculation code with PLINK codes and remove boost library. P-value calculated will be slightly different when compared to previous versions though the difference only occur at very small precision level, therefore should not affect the interpretation of data.
- Replace
--cov-header
by--cov-header
for consistency (w.r.t--pheno-col
)
Update (14/2/2018)
- Fix problem with --ld where when sample size of the reference and target is different, the clumping algorithm goes wrong
- Fix PRS calculation, now all score should systematically be 2 times the score generated before (because of not multiply the score by ploidy), this will have no effect on the R2 and p-value calculated
Update (8/2/2018)
- Fix problem with PRSet PRS where PRS does not clear out for new pathways
Note: Permutation for PRSet might still have memory error, we are currently trying to handle that
Update (26/1/2018)
- Fix memory problem with permutation (now we assume less memory are available, 1gb vs 10gb)
- Fix problem with seed not set
Update (25/1/2018)
- Fix problem with permutation
Support for Different Genetic Models
Update Log
- Now support different genetic models, including:
- add - Additive model, code as 0/1/2 (default)
- dom - Dominant model, code as 0/1/1
- rec - Recessive model, code as 0/0/1
- het - Heterozygous only model, code as 0/1/0
- Removed some of the unused functions
- Fix bug with permutation
- Add in fall back when ggplot or data.table cannot be install
- Will try to use the base graphics from R to draw the plots
- (Plots will have uglier legend when compared to those drawn by ggplot)
- Will try to use the base graphics from R to draw the plots
Bug fix
- Solved problem with BGEN file where PRSice didn't generate the list of valid SNPs
- Solved problem where the log file sometimes missed some error messages
20171109
- Fix problem with quantile plot error bar
- Can now directly execute the Rscript without using the Rscript call
20171111
- Fix problem when handling strand flipping for bgen file format
20171116
- Fix problem where PRSice doesn't allow no phenotype when using --no-regress
- Fix problem with keep-ambig
- Note: Due to my slow internet, I am not able to include the TOY dataset in this zip file
Memory Fix
Update Log
- Reduce memory usage, therefore reducing change of seeing the error message: Terminated something something uncaught exception
- Add PRSet and multi-target phenotype output graph
- Reorganize code such the PRSet will run slower but more memory efficient
- Add in fall back when data.table or ggplot2 cannot be install
Quick Bug fix
- log file name was not defined before seed output