Apparently the BigChain documentation https://genome.ucsc.edu/goldenPath/help/bigChain.html and https://genome.ucsc.edu/goldenPath/help/examples/bigChain.as is wrong about chainScore being a uint. Which explains a lot, because bedToBigBed fails with an integer overflowed, limit=4294967295 ... whenever trying to produce any kind of chains at scale. I'd previously hacked cactus-hal2chains to clamp this before bigBed conversion, but @diekhans has pointed me to the official bigChain.as which uses a double so this PR removes the clamp and switches to using this version.
Apparently the BigChain documentation https://genome.ucsc.edu/goldenPath/help/bigChain.html and https://genome.ucsc.edu/goldenPath/help/examples/bigChain.as is wrong about
chainScore
being auint
. Which explains a lot, becausebedToBigBed
fails with aninteger overflowed, limit=4294967295 ...
whenever trying to produce any kind of chains at scale. I'd previously hackedcactus-hal2chains
to clamp this before bigBed conversion, but @diekhans has pointed me to the official bigChain.as which uses adouble
so this PR removes the clamp and switches to using this version.