Troubleshooting

This page collects first-pass checks for common setup, alignment, and graph-review problems.

Troubleshooting Entry Points

Use these entry points when a command runs but the evidence, plot, or reviewed plan does not look right.

Symptom First check Deeper guide
chromo is not available Activate the installed environment Installation
MUMmer coords fail to parse or PAF rows disappear Check coords and PAF format assumptions Choosing PAF or MUMmer coords
Coords and PAF disagree Compare aligner settings and filters Choosing PAF or MUMmer coords
Edited FASTA was paired with old coords or PAF Re-align the exact edited FASTA Alignment evidence and the exact FASTA rule
Plot shows an old assembly Check plot provenance Alignment evidence and the exact FASTA rule
Plot is empty or sparse Check plot inputs and sequence names FASTA and evidence name matching
Audit or review status is confusing Compare status labels before changing sequence Reading ChromoSort audit tables
GFA nodes, overlays, or projections are missing Check graph and FASTA coordinate systems hifiasm unitig-to-contig projection
Gapfill candidates remain unresolved Review ambiguity and support evidence Graph-supported gap filling

chromo Command Not Found

Activate the environment that installed ChromoSort:

mamba activate chromosort
chromo --help

If you are using Pixi, run commands through Pixi:

pixi run help
pixi run test

MUMmer Coordinates Do Not Parse

Use show-coords -r -c -l on a filtered delta file. ChromoSort expects reference names, query names, reference/query coordinates, lengths, coverage, percent identity, and sequence names.

show-coords -r -c -l mummer/sample.filter > mummer/sample.coords

PAF Rows Are Missing

By default ChromoSort skips secondary PAF rows marked tp:A:S. Add --include-secondary-paf only when secondary rows are part of the review plan. For noisy alignments, also check --min-mapq.

For broader alignment-evidence choices, see Choosing PAF or MUMmer coords.

Coords And PAF Disagree

Small disagreements are expected. In soybean chromo fix testing, coords and PAF split counts differed by about 5-10%, while marginal split-contig sets differed by about 20-30%. ChromoSort normalizes both formats before decision logic, so first check aligner settings, minimap2 preset, -c --secondary=no, MAPQ filters, MUMmer delta-filter settings, row counts, and dot plots. Use chromo eval with long-read PAF, GFA, or GAF for stronger event evidence. The guide-level comparison is Choosing PAF or MUMmer coords.

Edited FASTA Does Not Match Old Alignments

After chromo sort writes ordered.fa, chromo clean writes clean.fa, chromo fix writes fixed.fa, chromo cut writes a cut FASTA, chromo manual apply writes a manual FASTA, or chromo scaffold writes scaffold records, re-align the edited FASTA before running another alignment-dependent command on that edited FASTA. Old raw assembly alignments usually do not describe the edited sequence names, membership, orientation, or coordinates.

It is fine to reuse raw coords or PAF for another decision about the same raw assembly, such as running chromo fix --assembly-fasta raw.fa --coords raw.coords after inspecting chromo sort --assembly-fasta raw.fa --coords raw.coords. It is not safe to run that same raw.coords against sample.ordered.fa or sample.fixed.fa.

For the full rule and stage-by-stage examples, see Alignment evidence and the exact FASTA rule.

Plot Shows The Old Assembly

chromo plot --assignments draws the original alignment rows and uses a chromo sort assignment table to order the query axis. This is useful for reviewing sort decisions. It does not validate ordered.fa, fixed.fa, or a manual FASTA unless the coords or PAF were generated from that exact FASTA. The same rule applies to clean.fa from chromo clean.

Use How to interpret dot plots for plot patterns, and Alignment evidence and the exact FASTA rule when the plotted FASTA pair is uncertain.

GFA Nodes Are Reported Missing

Graph-aware commands expect GFA segment names to match the assembly FASTA or the original contig names in the assignment report. If another tool renamed, polished, split, or scaffolded the FASTA after graph export, keep a name map or regenerate graph evidence for the renamed sequences.

See FASTA and evidence name matching and Assembly graph evidence for the guide-level view of these checks.

GFA Overlay Or Projection Is Empty

For hifiasm outputs, check whether your FASTA and GFA use the same coordinate system. Dot plots made from p_ctg.fa or hap*.p_ctg.fa are in contig coordinates, while p_utg.gfa, r_utg.gfa, and their .noseq.gfa equivalents are usually in unitig coordinates. ChromoSort can project unitigs onto contigs only when the GFA includes P path records or W walk records whose names match the contig FASTA records.

Run chromo graph-map first when you are unsure:

chromo graph-map \
  --ctg-fasta assembly.p_ctg.fa \
  --utg-gfa assembly.p_utg.noseq.gfa \
  --output-prefix review/sample.graphmap

If the warning table says paths are missing, the GFA may contain only S, L, and read-alignment A records. Those files are still useful for topology and junction context, but they cannot define contig-axis unitig intervals for chromo plot --gfa-overlay.

See hifiasm unitig-to-contig projection for the coordinate-system model behind this warning.

Gapfill Plan Is Ambiguous

chromo gapfill refuses to guess through ambiguous graph branches. Add GAF, Hi-C-like pair evidence, or reference-placement PAF when those evidence layers are trustworthy, then review the candidate-path table, chromo eval gapfill table, or HTML reviewer. Ties, weak support, conflicting evidence, missing sequence, invalid overlaps, and flank mismatches intentionally remain unresolved.

See Graph-supported gap filling for the candidate-path and risk-flag interpretation workflow.

Plots Are Empty Or Sparse

Check that the reference and assembly FASTA IDs match the alignment file exactly. If you applied strict filters upstream, try plotting the unfiltered or less strictly filtered coords/PAF file first, then add --min-segment-bp, --min-segment-idy, or --min-mapq only after confirming the expected rows are present.

If the plot is not empty but the pattern is hard to classify, use How to Interpret Dot Plots to compare it against common clean, reversed, chimeric, inversion, duplicate, gap, and repeat-like examples.

If the plot is empty because rows do not connect to the FASTA records, use FASTA and evidence name matching.