Statistical Genomics Foundry
Paper
domain/synteny

Lovell et al. 2022 — GENESPACE

1. Citation

Lovell JT, Sreedasyam A, Schranz ME, Wilson M, Carlson JW, Harkess A, Emms D, Goodstein DM, Schmutz J. 2022. “GENESPACE tracks regions of interest and gene copy number variation across multiple genomes.” eLife 11:e78526. Published 2022 Sep 9. doi:10.7554/eLife.78526. Software version described in the paper: GENESPACE v0.9.3.

2. Access note

Full text read via eLife HTML (https://elifesciences.org/articles/78526) and PMC (PMC9462846), fetched 2026-07-04. NOTE: the PMC id given in the invocation (PMC9648965) is WRONG — it resolves to an unrelated eukaryogenesis review; the correct PMC id for this paper is PMC9462846. Text was extracted through a summarizing fetch tool; the load-bearing quotes below were cross-confirmed across two independent fetches of Materials-and-methods passages, but I could not byte-verify every character against a raw PDF. Where a fact was not found, I record it ABSENT rather than infer.

3. Thesis (1 sentence)

GENESPACE integrates conserved gene order (synteny) with orthology to define the expected physical position of every gene across multiple genomes, yielding a synteny-constrained pan-genome annotation and a “riparian” cross-genome visualization.

4. Problem & context

“Leveraging information across multiple genomes remains a significant challenge in nearly all eukaryotic systems.” Existing gene-family / synteny tools handle either orthology or collinearity but not their intersection; GENESPACE targets that intersection, aimed “especially [at] polyploid, outbred, and other complex genomes.”

5. Method / approach (pipeline)

Ordered pipeline (Materials and methods):

  1. Inputs: “GENESPACE operates on gff3-formatted annotation files and accompanying peptide fasta files.” (one gff3 + one peptide FASTA per genome.)
  2. Homology: “GENESPACE calculates BLAST-like hits from DIAMOND2 and runs OrthoFinder to infer orthogroups and orthologues.” DIAMOND v2.0.8.146, OrthoFinder v2.5.4.
  3. Array collapse + rank order: genes are collapsed to “array representative” genes; “Gene rank order along the genome is recalculated on these ‘array representative’ genes.” Physical position for synteny = condensed gene rank-order position (not bp).
  4. Anchors / synteny: collinear hits found via MCScanX_h then buffered; “MCScanX_h (-s = blkSize, -m = nGaps) is called from R using condensed gene rank-order positions as the physical location, and those hits within collinear blocks are flagged as initial anchors.” “GENESPACE then extracts syntenic regions from the hits using a combination of graph- and cluster-based approaches, producing syntenic orthogroups for each unique (not reciprocal) pair of genomes.”
  5. Pan-genome annotation: the pangenome function decodes pairwise syntenic orthologs into a multi-genome pan-annotation against a chosen reference.
  6. Visualization: the plot_riparian function draws the riparian (braid) plot.

Exact default parameters (synteny function)

Verbatim: “the minimum number of unique hits within a syntenic block (‘blkSize’, default = 5), the maximum number of gaps within a block alignment (‘nGaps’, default = 5), and the radius around a syntenic anchor for a hit to be considered syntenic (‘synBuff’, default = 100).”

  • blkSize = 5 (minimum number of unique hits within a syntenic block)
  • nGaps = 5 (maximum gaps within a block alignment; passed to MCScanX_h -m)
  • synBuff = 100 (radius around a syntenic anchor for a hit to count as syntenic) No other synteny parameters with defaults are listed in Materials and methods.

R functions named in the paper

synteny, pangenome, plot_riparian. NOTE: the paper (v0.9.3) does NOT name init_genespace or run_genespace — those are a later (v1.x) driver API and are ABSENT from this source.

6. Key claims / findings

  • blkSize default is 5, but its unit is “unique hits within a syntenic block,” NOT “orthogroups.” A skill claiming “5 orthogroups (Lovell 2022 default)” mis-states the unit; the number 5 is correct, the unit is unique syntenic hits.
  • Anchor definition (verbatim): “potential anchor hits are defined as those where both the query and target are array representatives, the hit is not in maskTheseHits, the query scrRank ≤ nHits1, the target scrRank ≤ nHits2, and, optionally if onlyOgAnchors, both the query and target are in the same orthogroup.” → orthogroup-constrained anchors are OPTIONAL (onlyOgAnchors), not unconditional.
  • Software pins: OrthoFinder v2.5.4, DIAMOND v2.0.8.146, GENESPACE v0.9.3.
  • Synteny uses gene RANK ORDER (condensed, array-representative), not base-pair coords.
  • Syntenic orthogroups are built “for each unique (not reciprocal) pair of genomes.”
  • pangenome output: “a long-formatted text file, where each gene is given a reference genome syntenic position and chromosome, and flags”; plus “a wide-formatted data.table (R object) where each row is a pan-genome entry with positional information and each column is a list of genes by genome.”
  • Riparian anchoring: “Genomes are ordered vertically to minimize the number of translocations between each pairwise combination. Chromosomes are ordered horizontally to maximize synteny with the [reference] chromosomes.” “Chromosome segment sizes are scaled by the total number of genes in syntenic networks and positions of the braids are the gene order along the chromosome sequence.” Reference is user-chosen (in the sex- chromosome figure the reference is human, chromosomes [X, 1–22]).
  • Demonstrated across a wide taxonomic range: seven main genome pairs, 15 vertebrate genomes (sex-chromosome analysis), eight grass genomes, and 26 maize cultivars.

7. Load-bearing statements (VERBATIM — license permits reproduction; eLife CC BY)

  1. “the minimum number of unique hits within a syntenic block (‘blkSize’, default = 5), the maximum number of gaps within a block alignment (‘nGaps’, default = 5), and the radius around a syntenic anchor for a hit to be considered syntenic (‘synBuff’, default = 100).” — Materials and methods (synteny parameters).
  2. “GENESPACE operates on gff3-formatted annotation files and accompanying peptide fasta files.” — Materials and methods (inputs).
  3. “potential anchor hits are defined as those where both the query and target are array representatives, the hit is not in maskTheseHits, the query scrRank ≤ nHits1, the target scrRank ≤ nHits2, and, optionally if onlyOgAnchors, both the query and target are in the same orthogroup.” — Materials and methods (anchor definition).
  4. “MCScanX_h (-s = blkSize, -m = nGaps) is called from R using condensed gene rank-order positions as the physical location, and those hits within collinear blocks are flagged as initial anchors.” — Materials and methods (synteny).
  5. “The output is a long-formatted text file, where each gene is given a reference genome syntenic position and chromosome … pangenome also returns a wide-formatted data.table (R object) where each row is a pan-genome entry … and each column is a list of genes by genome.” — Materials and methods (pangenome).

8. Stated scope, assumptions, limitations

  • Assumes a single gff3 annotation + matching peptide FASTA per genome.
  • Core assumption on copy number: homologs should be effectively single copy within any syntenic region between a genome pair (array representatives collapse tandem arrays).
  • Positioned “especially [for] polyploid, outbred, and other complex genomes.”
  • No explicit maximum genome count and no plant-only restriction is stated; the paper itself applies GENESPACE to vertebrates and grasses, so it is not plant-limited.

9. Failure modes / invalidity patterns (referee-relevant)

  • Version drift: the pipeline pins external tools (OrthoFinder v2.5.4, DIAMOND v2.0.8.146). GENESPACE parses OrthoFinder output; a mismatched OrthoFinder version can break parsing. (Paper states the pinned version; it does not itself narrate the failure message — [summarizer-inferred] that a version mismatch is the mechanism.)
  • Input mismatch: gff3 and peptide FASTA gene IDs must correspond; mismatch breaks the annotation join. (Implied by the input contract; not spelled out as an error string.)
  • Mis-set blkSize/synBuff/nGaps change what counts as syntenic — too strict drops real blocks, too loose merges non-collinear regions. (Design consequence of the parameter definitions; the paper states the params, not the tuning failure explicitly.)
  • Treating blkSize’s unit as orthogroups rather than “unique hits” would mis-document the threshold.

10. What the source does NOT address (confident silences)

  • Exact OUTPUT FILENAMES: syntenicHits.txt, pangenes.txt, pangenome.txt are NOT named in the paper — ABSENT. The paper describes output content/format (long text file + wide data.table), not on-disk filenames.
  • The driver functions init_genespace / run_genespace are ABSENT (paper = v0.9.3, functions synteny / pangenome / plot_riparian).
  • No stated hard upper bound on number of genomes; no “>30 genomes” claim.
  • No R-version or GENESPACE ≥1.4.0 requirement (paper predates that; it is v0.9.3).

11. Open questions / ambiguities

  • Whether orthogroup-constrained anchoring (onlyOgAnchors) is on by default is not clearly stated — the anchor definition marks it “optionally.”
  • Mapping between the paper’s v0.9.3 API (synteny, pangenome) and the later init_genespace / run_genespace driver API is not covered by this source.

12. Guidance answers

  1. Minimum syntenic block size / is default 5? YES — blkSize default = 5, but the unit is “the minimum number of unique HITS within a syntenic block,” not orthogroups. The number backs “default = 5”; the “orthogroups” unit is over-claimed. (synBuff = 100, nGaps = 5 are the companion synteny params.)
  2. Pipeline / anchor definition. OrthoFinder (+DIAMOND2) → array-representative rank order → MCScanX_h collinear anchors (-s = blkSize, -m = nGaps) → synBuff-buffered syntenic hits → syntenic orthogroups per unique genome pair → pangenome → riparian. Anchor definition quoted verbatim in §7 #3 (array representatives, not in maskTheseHits, scrRank ≤ nHits, optional onlyOgAnchors same-orthogroup).
  3. OrthoFinder version. “OrthoFinder v2.5.4” (DIAMOND v2.0.8.146). Backs a version-mismatch failure mode.
  4. Input format. “gff3-formatted annotation files and accompanying peptide fasta files” (§7 #2). No explicit “.bed” conversion is named in the paper text I read; genes are internally recast to condensed gene RANK-ORDER positions (array representatives).
  5. Riparian + pan-gene outputs. Riparian: braids of syntenic gene order, genomes ordered vertically to minimize translocations, chromosomes horizontally to maximize synteny with a chosen reference (see §6). Pan-gene output: long-format text file (each gene → reference syntenic position/chromosome + flags) plus wide data.table (row = pan-genome entry, column = genes by genome). Filenames syntenicHits.txt / pangenes.txt are ABSENT from the paper.
  6. Plant-specific assumptions / scaling >30 genomes / non-plant clades. No plant-only restriction and no “>30 genome” or hard genome-count limit is stated. The paper explicitly runs on vertebrates and grasses; it frames its advantage as “especially in polyploid, outbred, and other complex genomes.” So the stated boundary is genome COMPLEXITY, not clade.
  7. R function names + key args. Paper names synteny (args include blkSize, nGaps, synBuff, and flags maskTheseHits, onlyOgAnchors), pangenome, and plot_riparian. init_genespace and run_genespace are NOT in this source (they are a later-version API).