read mapping single samples

[jaudall1]

I'm trying to map single samples and this function of phg seems fairly buggy. I'm doing single samples at a time to map different samples in parallel, instead of serially from a single keyfile (e.g. walltime limits etc.).

my command (it is still annoying that it'll crash if the directory doesn't exist):
mkdir -p output/read_mappings/BPS1240_t40; export JAVA_OPTS='-Xmx270g'; ../bin/phg map-reads --threads 34 --hvcf-dir output/vcf_files --index output/index_files/phg27.fmd --read-files data/fastq_output/SRR6311492_1.fastq.gz,data/fastq_output/SRR6311492_2.fastq.gz --output-dir output/read_mappings/BPS1240_t40

From my output from one of the files:
[main] INFO net.maizegenetics.phgv2.api.HaplotypeGraph 2025-08-01 09:24:39,798: rangeToSampleToChecksum: 124921 x 55
[main] INFO net.maizegenetics.phgv2.api.HaplotypeGraph 2025-08-01 09:24:39,798: numOfSamples: 55
[main] INFO net.maizegenetics.phgv2.api.HaplotypeGraph 2025-08-01 09:24:39,799: numOfRanges: 124921
[main] INFO net.maizegenetics.phgv2.pathing.ropebwt.MapReads 2025-08-01 09:24:41,866: Mapping reads to pangenome
[main] INFO net.maizegenetics.phgv2.pathing.ropebwt.MapReads 2025-08-01 09:24:41,867: Mapping reads in data/fastq_output/SRR6311492_1.fastq.gz to output/index_files/phg27.fmd
[main] INFO net.maizegenetics.phgv2.pathing.ropebwt.MapReads 2025-08-01 11:17:36,735: Writing read mapping to output/read_mappings/BPS1240_t40/SRR6311492_1_1_readMapping.txt
[main] INFO net.maizegenetics.phgv2.pathing.ropebwt.MapReads 2025-08-01 11:17:37,013: Mapping reads in data/fastq_output/SRR6311492_2.fastq.gz to output/index_files/phg27.fmd

And yet the output file only has the header in it.
-rw-rw-r-- 1 mefftrrfrf proj-cotton_genomics 89 Aug 1 11:17 output/read_mappings/BPS1240_t40/SRR6311492_1_1_readMapping.txt

there aren't any other errors or warnings. What is going on? I've tried many different iterations, but always the same result. Are there temp files that need to be deleted? I have been testing on the same samples. Is there an entry in the database that flags the creation of the read_mapping output?

BTW - the website says "If you do not want to create a keyfile and only have one sample ...", but using the --read-files parameter phg creates a keyfile anyway. Is there a way to toggle that off??

[mtkelleher]

How long are your reads? I may have had a similar issue. If you run phg map-reads --help you will see some of the default options for mapping reads. Some of the parameters for the ropebwt3 method aren't yet mentioned in the documentation on https://phg.maizegenetics.net/imputation_ropebwt/#read-mapping.

The default --min-mem-length is 148, which is longer than any of the reads I was mapping, so I changed this to a lower number. The --max-start by default is 0, which I believe means that each match has to start at the first position, but I set this to > the length of my reads so that the match could start anywhere. Similarly, I changed the --min-end to 0 so that this parameter was not preventing reads from mapping.

[jaudall1]

Thanks! I'll give it go. I was thinking in a samtools/bwt world where defaults are often good for Illumina reads.