gradebins.sh error with gtdb

[michoug]

Hi
When running

gradebins.sh ref=myo.fa in=bin_dir gff=my.gff gtdb=gtdbtk_output_dir/ report=out.tsv clade=f

I got this error

Loading myo.gff
Loading tax tree from /scratch/gmichoud/HF_bact/.pixi/envs/default/opt/bbmap-39.81-1/resources//tree.taxtree.gz
6.910 seconds.
Made size map.
Exception in thread "main" java.lang.NullPointerException: Cannot read the array length because "bytes" is null
        at java.base/java.lang.String.<init>(String.java:546)
        at parse.LineParser1.parseString(LineParser1.java:233)
        at bin.GTDBLine.<init>(GTDBLine.java:32)
        at bin.GradeBins.loadGTDBFile(GradeBins.java:1263)
        at bin.GradeBins.loadGTDBDir(GradeBins.java:1238)
        at bin.GradeBins.process(GradeBins.java:346)
        at bin.GradeBins.main(GradeBins.java:66)

An AI agent suggested that the error may be :

Root cause identified via bytecode inspection: the classification field is parsed using the semicolon-delimited LineParser1 (lpsemi) instead of the tab-delimited one (lptab), and lpsemi is never .set() with any bytes before use.

The single-arg constructor GTDBLine(byte[]) builds two parsers:

new LineParser1(9)    // tab-delimited, .set() with the line bytes  → lptab
new LineParser1(59)   // semicolon-delimited, never .set() with anything → lpsemi

The two-arg constructor GTDBLine(LineParser1, LineParser1) then does:

name           = lptab.parseString(0);   // correct
classification = lpsemi.parseString(1);  // BUG — should be lptab.parseString(1)

Since lpsemi was never given any bytes via .set(), its internal bytes array is null, and calling parseString(1) on it triggers the NPE inside String's constructor — unconditionally, on every input line, regardless of content.

Cheers
Greg

[btpsj]

Have you tried this with the latest version of BBTools? It looks like you are using 39.81. At the time of this issue, July 16, the most recent version was 39.97, and between those two versions there have been several bug hunt fixes added.