Result disagrees with ucsc's liftover

[carbocation]

Example go program:

package main

import (
	"fmt"

	glo "github.com/blachlylab/GLO"
)

func main() {
	liftover := new(glo.LiftOver)
	liftover.Init()

	liftover.LoadChainFile("hg19", "hg38", "hg19ToHg38.over.chain")

	target := glo.NewChainInterval("chr1", 120887250, 120887251)
	response := liftover.Lift("hg19", "hg38", target)
	fmt.Println(response)
}

Result:

• [chr1:143924375-143924377]

Meanwhile, providing the same range to UCSC's liftover (chr1 120887250 120887251) yields:

• chr1 143924010 143924011

[jblachly]

I think I may have known about this. I fixed it when I rewrote it in Dlang (https://github.com/blachlylab/swiftover). I'll have to reach back into my memory to remember the reason this happens so that you can fix it in your fork.

[jblachly]

However I do recall noting some circumstances where I disagreed (subjectively) with UCSC's results in cases of discontiguous chains; i.e., I believe that UCSC's result was incorrect.