Large hyperpartition groups missing from groups.agp after scaffolding (polyploid CiFi dataset)

[DR-Omics]

Hi,
Thanks for the great tool.
I ran the full cphasing pipeline using CiFi data (highly polyploid plant genome). Mapped 3 CiFi datasets (3 fastq's) of same library separately to the contig-level assembly and merged them into one porec.gz and then ran cphasing pipeline on the merged dataset

cphasing mapper $REF $LIB2 --mm2-params "-x map-hifi" -t $THREADS -o $WRKDIR/CiFi_lib2_Mar2026
cphasing mapper $REF $LIB2_FULL --mm2-params "-x map-hifi" -t $THREADS -o $WRKDIR/CiFi_lib2_Full
cphasing mapper $REF $LIB2_PART --mm2-params "-x map-hifi" -t $THREADS -o $WRKDIR/CiFi_lib2_PART
# Merge porec.gz
cphasing porec-merge $WRKDIR/CiFi_lib2*.porec.gz -o $WRKDIR/CiFi_lib2.merged.porec.gz
# Cphasing pipeline
cphasing pipeline \
  -f "$REF" \
  -pct "$WRKDIR/CiFi_lib2.merged.porec.gz" \
  -t $THREADS \
  -n 10:0 \
  -o "$OUTDIR/CiFi_Lib2_mergedLibs_cphasing_n10_0" \
  --min-contacts 1 \
  --min-scaffold-length 10000 \
  --mm2-params "-x map-hifi" \
  -p CATG

Hyperpartition completed successfully and produced large subgroup partitions such as:

1g1   99,754,319
1g2   131,913,392
1g3   92,607,939
1g4   102,777,481
1g5   100,700,259
...
1g101 26,898
2g1 148,030,961
...
10g1  63,671,421
10g2  59,608,904
...
10g10 56,175

with total partitioned size:
Total 5,454,328,446
However, after scaffolding:
groups.agp seems to start only from smaller groups like Chr01g8, while large groups (Chr01g1–g7) are absent.
Same issue occurs across all 10 chromosomes.Example from AGP stats:

Chr01g8   4   38742592
Chr01g11  2   127880
Chr01g12  1   116322
...
Chr08g8   1   272317
Chr08g9   1   71971
Chr08g10  1   45446
...
Chr09g4   1   22676471
Chr09g6   2   22266383
...
Chr10g8 1       100902
Chr10g10        1       56175
Total number of contigs:        3411
Total length of contigs (bp):   5533696479
Total number of anchored contigs:       170
Total length of anchored contigs (bp):  171330149
Total length of chromosome level assembly (bp): 171331049
Number of unanchored contigs:   3241
Length of unanchored contigs (bp):      5362366330
Anchor rate (%):        3.10

Question: Why are the major hyperpartitioned groups excluded during scaffolding? Also, in plots, the chromosome subgroup labelling is collapsed into Chr01, Chr02, etc., Please see attached plot. Any advice appreciated.

Thank you
Dhanu

[wangyibin]

Hi,

Could you please provide the following files for debugging?

• output_dir/cphasing_xxx.log
• 3.hyperpartition/output.clusters.txt
• $REF.contigsizes

These files will help us better understand the issue. Thanks!

[DR-Omics]

Thanks for the response wangybin. Attached all three files.

cphasing_05-15_18_20_06.log.txt

output.clusters.txt

Reference_50X.asm.p_ctg.contigsizes.txt

[wangyibin]

Thanks for providing the files.

I initially suspect that the issue may occur during the ordering and orientation step for each group in 4.scaffolding. It looks like some expected output files were not generated, possibly because the per-group scaffolding process failed.

Could you also try rerunning the scaffolding step with --keep-temp enabled in 4.scaffolding/scaffolding.cmd.sh, and possibly with fewer threads?

Also, please provide the log files from:

• 4.scaffolding/logs
• scaffolding_tmp*/logs/*

These logs should help determine whether allhic optimize failed for any group.

Thanks!

[DR-Omics]

Thanks for the suggestion. I reran the scaffolding step with --keep-temp enabled and reduced threads to -t 8.
For the large groups that are missing from groups.agp (for example, Chr01g1), only the following files are generated:
Chr01g1.clm
Chr01g1.txt
However, no .tour or .tour.sav files are produced.
I do see .tour files for subgroups that appear in groups.agp file.
I have attached the requested log files from:
4.scaffolding/logs
scaffolding_tmp*/logs/*
The repeated errors in HapHiC_sort.log are:
generator object is not subscriptable
along with:
allhic optimize ... returned non-zero exit status 1
Is this something with software compatibility issue?

splitclm.log
HapHiC_sort.log

Chr01g48.allhic_stdout.log
Chr01g16.allhic_stdout.log
Chr01g48.allhic_stderr.log
Chr01g11.allhic_stdout.log
Chr01g16.allhic_stderr.log
Chr01g11.allhic_stderr.log
Chr09g6.allhic_stdout.log
Chr01g17.allhic_stdout.log
Chr01g13.allhic_stdout.log
Chr01g17.allhic_stderr.log
Chr01g8.allhic_stdout.log
Chr01g13.allhic_stderr.log
Chr01g8.allhic_stderr.log
Chr09g6.allhic_stderr.log

Thanks again for your help.

[wangyibin]

Hi,

Sorry for the inconvenience caused by this bug.

We have identified the issue and are working on a fix. For now, could you please try replacing the following file in your local C-Phasing installation?
HapHiC_sort.py

CPhasing/cphasing/algorithms/HapHiC_sort.py

Alternatively, you can also try modifying the parameter in 4.scaffolding/scaffolding.cmd.sh from -m precision to -m allhic

Thanks again for helping us debug this issue!

[DR-Omics]

Thank you for identifying the issue and sharing the fix.
I replaced the HapHiC_sort.py file in my local C-Phasing installation and reran the scaffolding step with -m allhic. The anchor rate improved substantially from ~3–4% to 98.55%, and I can now see all major groups properly included in groups.agp. I’ve attached the updated groups.agp.stat screenshot below for your reference.

Total number of contigs:        3411
Total length of contigs (bp):   5533696479
Total number of anchored contigs:       1352
Total length of anchored contigs (bp):  5453207263
Total length of chromosome level assembly (bp): 5453327763
Number of unanchored contigs:   2059
Length of unanchored contigs (bp):      80489216
Anchor rate (%):        98.55

Thanks again for the quick fix and support.
Dhanu