Version 1.0.8

• Changed package structure to fix pip issue.

• Improved testing framework, and refactoring.

• Enabled testing when installed via pip, using ldpred-unittest commands

• Fixed a bug in LDpred that could improve convergence for gibbs.

• Other minor improvements

Version 1.0.7

• Improved handling of variants with p-values rounded down to 0.

• Added options to enable effect estimate inference from standard errors (--z-from-se) in coordination step.

• Fixed a serious bug that caused sample sizes in summary stats file not always be used correctly when provided.

• LDpred gibbs can now handle differing sample sizes per variant effects, if they are parsed in summary stats. This can improves convergence and accuracy subtantially, but slows the gibbs sampler (approx 5x). It can be avoided by setting --N flag.

• If the heritability is not provided when running LDpred gibbs and inf it now estimates it separately for each chromosome, which can improve accuracy if summary stats sample size is large. To avoid this behavior, either set --h2 flag, or use --use-gw-h2 flag.

• Improved reporting and other things.

Version 1.0.6

• Fixed #68
• Improved warnings and error messaging
• Added new flags for sum stats custom format, making it more flexible.

Version 1.0.5

• Fixed the --only-hm3 flag #65
• Fixed X chromosome issue #66
• #63
• other minor improvements

Version 1.0.4

• Fixed #64, i.e. problems with the --only-score flag
• Fixed #61
• Fixed #60
• Improved progress percentage for scoring and summary stats loading
• plus other minor reporting improvements

Version 1.0.3

• Fixed the --max-freq-discrep bug. This flag seems to help boost prediction accuracy significantly on some data. Is set by default to 0.1
• Other minor bugs.
• Improved reporting.

Version 1.0.2

• Fixed critical bug #58
• Fixed critical bug #57
• Simplified/Improved terminal output reporting
• Simplified resulting score files
• Added --only-score flag
• other minor improvements

Version 1.0.1

• Now the coordination step always compares the summary stats nucleotides and allele frequencies with the LD reference data when matching them.
• Minor bug fixes.

Version 1.0.0

Finally, here is the version 1.0.0 release.

It includes lots of improvements.

• New and more strucutred parameter and argument handling
• More clearer usage description
• A new Q and A wiki
• Switched to python3
• Improved summary statistics handling
• Improved ld information writing and storing
• lots of small bug fixes.

Version 0.999

The final BETA release.