A new start

Update Log

• Now incorporate the --id-delim parameter which might be useful for BGEN input. You can change the delimitor between the FID and IID information according to your BGEN file (NOTE: If you don't have the FID, and the BGEN file contain both the FID and IID, this will not help. You will need a proper sample file or a phenotype file with both the FID and IID information)
• --memory parameter should now work for both c++ and Rscript
• (Ghost Edit: Recompiled the Linux executable for it to work on other machines)

Side Note:

• 2.1.10 contains some of the fix for --id-delim but didn't seems to solve the problem. Therefore I don't make a release for it.
• I have now moved to Mount Sinai in New York, thus support will now follow the EDT

Competitive P-value Fixed

Update Log

• Fixed problem with the output of Competitive p-value
  • Competitive P-value results should now be included in the .summary file
• Linux compilation version should work on most linux server

BGEN completed

Update Log

• BGEN related function has now gone through extensive checking
• Hard coding now behave as expected. Algorithm now better follow the PLINK PRS calculation code
• Competitive P value should now be reported in the summary file correctly

PRSet Bug Fix

Update Log

• Fix problem with PRSet best score output when --set-perm isn't used
• Fix problem with --set-perm not possible
• Fix problem with --missing CENTRE with BGEN
• Fix missingness imputation problem with BGEN

Minor Bug Fix

Update Log

• Fix missing genotype imputation for binary plink genotype format
• Fix missing genotype imputation for binary gen format
• Fix multi-set-plot device error
• Now include a version for 32bit Windows
• Fix program not running on Windows

Note

With this bug fix, the PRS generated from PRSice should now be almost identical to those generated from PLINK

Minor Bug Fix

Update Log

• Fix log output related to bgen
• Fix log output related to --missing
• Fix bug in Rscript that prevent generation of strata plot

Note

• This is a re-upload of 2.1.5 as previous upload was associated with an incorrect git
• As the development of PRSice seems to become stable, we will now try to reduce the patching of PRSice to at most once per month unless there are bugs that affects the performance of the program.

PRSet is live

Update Log

This is more of a clean up release as 2.1.3.beta is getting messy

I am also making this our first official release as the number of bug seems to die down a bit. Hopefully, from 2.1.4 to 2.1.5, I should have completely finished the implementation of all relevant test cases

• Fix problem related to Rscript (Pheno not found)
• Fix problem related to --missing
• Set base PRS analysis tool: PRSet is now official live. Please refer to our documentation for more information

Speedy Covariate

Update Log

• Fix problem with .valid file output
• Re-structure code to be more cache friendly - Should lead to huge speed up

Add new features:

• --allow-inter: allow generation of large intermediate file when using BGEN as input. This allow PRSice to avoid repeat decompression when performing clumping, thus should speed up BGEN clumping.
• --cov-factor: allow user to specify categorical covariates. By specifying the column name of the categorical covariate in --cov-factor, PRSice will automatically generate the dummy variable and use it as the covariate
• --non-cumulate: allow non-cumulative calculation of PRS. i.e. When specified, will reset PRS to 0 for each threshold before calculation
• --snp-set and --snp-sets: Allow the use of SNP set(s) as an input for PRSet
• --x-range: Exclusion range, allow users to exclude range of genome from PRS altogether (e.g. MHC)

2018-07-23

• Fix no valid PRS problem with multi-phenotype input
• Fix problem with --ld

2018-08-03

• Fix bar-levels not honored by Rscript
• Fix problem with covariates in Rscript
• Fix all-score output rounding error and all-score output error
• Add warning message when column not found
• Fix problem with covariate reading

2018-08-10

• Fix problem with all-score output when no-regress is used
• Fix problem with incorrect Set membership for SNPs

2018-08-16

• Fix problem with --ld (You should use this version if you have previously use --ld)
• Fix problem with multi-bed input
• Allow users to not provide a background

2018-08-21

• Fix problem with multi-bed input

2018-09-05

This is supposed to be 2.1.4, but as I am have not finish something critical for 2.1.4 (basically, the full unit test), I will release this as a temporary interim update under 2.1.3. This is the product from a comprehensive code auditing. I have also build in some unit tests. Once all unit tests are implemented, PRSice should be ready for an official release and should be less buggy.

Below is the bug fix log:

• Fix problem with not identifying covariate in the covariate file (forgot to remove new line character)
• Fix segmentation fault when samples contain missing covariates (forgot to update the size of independent variable matrix)
• Fix malform bed file error (did not update the if stream status)
• Fix various off by one error in defining gene sets (captured by our unit tests)
• Fix a number of bugs related to bgen input (e.g. PRS is systematically 3 times smaller)
• Fix problem when handling missing genotype
• --x-range should now function properly even when unsorted listed input is provided (Use binary search instead)
• Quantile plots should now output OR for binary traits
• Fix problem where PRSice mistake quantitative trait is invalid with only 1 input

2019-09-19

I am still terribly behind in working towards 2.1.4 and because of up coming conferences, I don't expect that to be done soon (the only thing hold us back from 2.1.4 is me not completing all unit tests). Here are some bug fixes regarding BGEN

• Fix problem with unable to seek error (that's due to a careless mistake in an if case)
• Fix problem with the use of --allow-inter where you will see a not able to open file error message (because I have some logic mistakes)
• --allow-inter should now correctly work when reference panel isn't provided
• Fix problem with missingness calculation in dosage score (won't notice, because it is impossible to reach this point with previous releases)
• Fix glibc error (this is an out of bound error in genotype read)
• Add an more informative error message when std::bad_alloc is observed (whenever you see std::bad_alloc, that is almost always because there isn't enough memory space)

Quantile Breakthrough

Update Log

• Fix various issues related to BGEN
• Re-coded the Rscript for better readability and allow for easier management of codes
• Reformat most output files to reduce the number of files generated when PRSet is performed
• Introducing --quant-break which allows for uneven quantile distribution
• Now also output data used for quantile plot to a txt file

2018-5-15

• Fix quantile plot ordering problem

2018-5-31

• Fix chromosome number problem
• Fix window line ending problem
• Fix no regress all score output

2018-6-1

• Minor fix of the Rscript

2018-6-5

• Fix window output distortion problem

Quality of Life Update

Please use the 4th April or later version!

Update Log

10th April Update

• Fix problem with --all-score (also --no-regress) output

9th April Update

• Fix problem when using bgen together with a bgen ld reference.

4th April Update

• 29th March version is bugged. Whenever you use permutation, it will crash. This problem has now been fixed. ( I promise I will always run my test run sets before updating the release from now on)

29th March Update

This serves as a intermediate update, including all the bug fix since v2.1.0.beta.

Some notable changes includes:

1. Drastically reduce the memory requested by clumping, thus avoid crashing in cluster environment where memory is monitored
2. Drastically reduce the memory usage when --ld is used. There is no point including two copy of the SNP information
3. Proper implementation of the exclusion of non-founders. Previous versions failed to exclude non-founder from the regression model (but does exclude them from the ld calculation)
4. Introduce better memory control for permutation. PRSice will no longer crush during permutation due to memory problem
5. Slight improvement in capturing error in input

To Do List

1. Speed up BGEN clumping? (If possible)
2. Implement Set base permutation (for competitive testing)
3. Allow range to be added to set boundaries