Bioconductor Release 3.16

CHANGES IN VERSION 1.34.0

• fix the compiler issue on Mac M1 chip
• fix the C stack error in RStudio

Bioconductor Release 3.15

CHANGES IN VERSION 1.32.0

• fix the issue on Win32 because of using deprecated tbb::task_scheduler_init

CHANGES IN VERSION 1.30.2

• require GCC >= v8.0 for compiling the AVX-512VPOPCNTDQ intrinsics
• fix hlaGDS2Geno() when loading a SeqArray GDS file

CHANGES IN VERSION 1.30.0

• add the support of Intel AVX-512VPOPCNTDQ intrinsics (faster than AVX512BW)

Bioconductor Release v3.13

CHANGES IN VERSION 1.28.0

• hlaPredict() returns the dosage of HLA alleles when type="response+dosage", and hlaPredict() returns the best guess and dosages by default
• a new option "Pos+Allele" in hlaPredict(), hlaGenoCombine(), hlaGenoSwitchStrand(), hlaSNPID() and hlaCheckSNPs() for matching genotypes by positions, reference and alternative alleles; it is particularly useful when the training and test set are both matched to the same reference genome, e.g., 1000 Genomes Project
• hlaGDS2Geno() supports SeqArray GDS files
• a new option 'maf' in hlaAttrBagging() and hlaParallelAttrBagging()
• 'pos.start' and 'pos.end' are replaced by 'pos.mid' in hlaFlankingSNP() and hlaGenoSubsetFlank()
• new function hlaAlleleToVCF() for converting the imputed HLA classical alleles to a VCF file

CHANGES IN VERSION 1.26.1

• the hlaAttrBagging object can be removed in garbage collection without calling hlaClose()
• enable internal GPU API
• improved multithreaded performance compared with v1.26.0

CHANGES IN VERSION 1.26.0

• users can interrupt the model building in an interactive R session
• remove hlaErrMsg() since it is never used
• a new option 'nthread' in hlaAttrBagging() as a complement to hlaParallelAttrBagging()
• kernel version 1.5: generates the same training model as v1.4, but 2-6x faster, by taking advantage of Intel AVX, AVX2 and AVX512 intrinsics
• new function hlaSetKernelTarget() to automatically select the CPU target the algorithm is optimized for

Bioconductor Release 3.6

• Kernel Version: v1.4
• modify the kernel to support the GPU extension
• The kernel v1.4 outputs exactly the same parameter estimates as v1.3, and the model training with v1.4 is ~1.2 times faster than v1.3.

Bioconductor Package

• Kernel Version: v1.3
• optimize the calculation of hamming distance using SSE2 and hardware POPCNT instructions if available
• hardware POPCNT: 2.4x speedup for large-scale data, compared to the implementation in v1.2.4
• SSE2 popcount implementation without hardware POPCNT: 1.5x speedup for large-scale data, compared to the implementation in v1.2.4

The last version of HIBAG on R-Forge

The R package HIBAG will be submitted to Bioconductor and it is the last version on R-Forge/CRAN before HIBAG becomes a Bioconductor package formally.

Changes in 1.2.4-1.2.5

• Bug fixes on big-endian machines
• add a new function 'hlaGDS2Geno' to support SNP GDS files (in the R package SNPRelate)
• improve the display of 'summary.hlaAlleleClass'
• fix warnings in R-3.1.1 CMD check

Release HIBAG_1.2.3

Changes in 1.2.2-1.2.3:

• avoid missing values in 'hlaFlankingSNP'
• add more gene information in 'hlaLociInfo'
• make the C++ function 'HIBAG_SortAlleleStr' compatible with the clang++ compiler (Mac Mavericks)
• improve the function 'hlaReport'
• improve document