changelog.md

In SeojinHwang/scadsum: nonconvex penalty with summary statistics and a reference panel

V0.2.0

• Initial release

v0.2.1

• Use a c++ function for counting lines instead of bash wc
• Corrects bug in lassosum.pipeline() when calling some functions in data.table

v0.2.2

• includes some additional objects in the output of validate.lassosum.pipeline() and pseudovalidate.lassosum.pipeline()
• Allows the validate.function in validate.lassosum.pipeline() to be given as a character rather than a function.

v0.2.3

• best.beta is output in validate.lassosum.pipeline() and pseudovalidate.lassosum.pipeline()
• fixes bug when grouping blocks that are too small

v0.2.4

• fixes bug when matching variable names in matchpos()
• Support for parallel processing in the pgs() function, (and therefore validate.lassosum.pipeline() and pseudovalidate.lassosum.pipeline())
• Allows summary statistics to be matched by SNP id as well as chromosome/position
• fixes bug when the bfile is too large (due to integer overflow in c++)
• adds a user-defined limit to the size of the bfile to discourage the use of large reference panels
• cpp functions are now exported (except openPlinkBinaryFile())
• includes this changelog file
• fixes bug in plotting results from validate.lassosum.pipeline() and pseudovalidate.lassosum.pipeline()
• fixes bug in displaying "Running lassosum with s=1..." message.
• renames hg19 LD region files
• adds hg38 LD region files (lifted over from the hg19 ones)

v0.2.5

• Bug fix for lassosum.pipeline() in parsing LDblocks when given as a vector of factors
• Bug fix for pgs() when extract is specified

v0.3.0

• Adds function for cross-prediction link
• Allows multiple bfiles in many functions

v0.3.1

• Adds cp.meta() for meta-analysing external summary statistics with results from cp.plink.linear()
• cp.plink.linear() will weight correlation by sample size when averaging across folds
• A number of unrelated functions are now hidden.
• Improved documentation for cp functions

v0.3.2

• fixes bug in cp.plink.linear covar option
• fixes bug when specifying keep
• adds write.table2()

v0.4.0

• Removes support for cross-prediction.
• Cross-prediction moved to a separate package
• recast validate.lassosum.pipeline(), pseudovalidate.lassosum.pipeline(), and pgs as S3 methods
• Adds option for including covariates in validate.lassosum.pipelin()
• bug fix in matchpos()
• Introduces c++ function for pgs() with sparse matrix. Used as default.
• p2cor() will automatically code to NA those correlations with relatively small n.
• Always uses data.table::fread() for reading files

v0.4.1

• Bug fix for matchpos()
• pgs() stops if keep or extract not in the same order as in .bim/.fam file

v0.4.2

• Allows covar and pheno to take data.frames or files as input.
• You no longer have to re-enter test.bfile if keep or remove is used in validate.lassosum.pipeline() or pseudovalidate.lassosum.pipeline()
• Adds splitvalidate() for split-validation.
• Adds subset.lassosum.pipeline() for applying validated PGS to new data.
• Allows pheno to be constant or NA if only one PGS is calculated in validate.lassosum.pipeline().
• plot.validate.lassosum() will abandon if there are no finite values.
• A standalone version of lassosum is available

v0.4.3

• LDblocks does not need to be specified when s < 1
• Better tracing in SD calculation
• Bug fix in validate.lassosum.pipeline()
• Allows multiple test.bfile in validate.lassosum.pipeline()

v0.4.4

• Bug fix
• Better documentation for Mac users

v0.4.5

• [Bug fix] standalone version used to ignore --nthreads option when running validate.lassosum.pipeline
• [Bug fix] The pheno and covar option didn't use to like data.table and required data.frame
• [Bug fix] Pull request #20
• Adds seed option in standalone version

SeojinHwang/scadsum documentation built on June 30, 2023, 10:52 p.m.