SyRI (Goel et al. 2019) — source note

1. Citation

Goel M, Sun H, Jiao W-B, Schneeberger K. 2019. “SyRI: finding genomic rearrangements and local sequence differences from whole-genome assemblies.” Genome Biology 20:277. doi:10.1186/s13059-019-1911-0.

PMID 31842948; PMCID PMC6913012 (open access).
License: Genome Biology is CC BY 4.0 → short verbatim load-bearing quotes permitted.

PMC-ID correction: the invocation gave PMC6912966; that ID resolves to an unrelated cardiomyopathy review (van Waning et al., JAHA), not SyRI. The correct SyRI PMCID (via Europe PMC DOI lookup) is PMC6913012. This note summarizes PMC6913012.

2. Access note

Read the full-text XML via Europe PMC REST (.../PMC6913012/fullTextXML), including figure legends for Figs 1–3. Supplementary files (Additional file 2, e.g. Table S3 with exact aligner command lines) were NOT retrieved — those specifics are marked ABSENT/ “in supplement” below. No paywall on the main article.

3. Thesis (1 sentence)

SyRI is a pairwise whole-genome-comparison tool for chromosome-level assemblies that first identifies structural rearrangements (inversions, translocations, transpositions, duplications) and then the local sequence differences within them, distinguishing whether each difference resides in a syntenic vs a rearranged region — a distinction that matters because rearranged regions are inherited differently.

4. Problem & context

Existing methods annotate SNP-to-large-indel differences accurately but “do not unravel the full complexity of structural rearrangements, including inversions, translocations, and duplications, where highly similar sequence changes in location, orientation, or copy number” (Abstract). SyRI’s premise: with two chromosome-level assemblies available, one can identify ALL differences (structural + sequence) and their breakpoints, rather than inferring SVs indirectly from reads.

5. Method / approach

Input: “SyRI expects whole-genome alignments (WGA) as input.” A WGA = a set of local alignments, each connecting a region in one genome to a region in the other.

Aligner used: MUMmer3 toolbox (nucmer). “Nucmer was run with --maxmatch to get all alignments between two genomes and also included -c, -b, and -l parameters” (values chosen to balance resolution/runtime; full commands in Additional file 2: Table S3 — not retrieved here). nucmer output is post-processed with delta-filter and show-coords into tab-delimited coords.
“other alignment tools like minimap2 [18] can be used as well.” Paper does not, in the retrieved main text, document direct SAM/BAM ingestion or a -F/format flag → see §9/§12.

Three-step algorithm (Fig 2):

Step 1 — syntenic path. “SyRI identifies the highest scoring syntenic path between the corresponding genomes.” “The syntenic path represents the longest set of non-rearranged regions between two genomes.” Implemented via dynamic programming from nodes at one chromosome end to nodes at the other end. “Since all non-syntenic regions are rearranged by definition, identifying syntenic regions identifies rearranged regions at the same time.”
Step 2 — classify rearrangements. Remaining (off-path) alignments split into structural rearrangements vs redundant alignments. “Structural rearrangements (green alignments) are classified into inversions, transpositions, and duplications, and finally inter-chromosomal rearrangements.” “Low scoring candidates and those that are overlapping with syntenic or inverted regions are filtered out.”
Step 3 — local variations. “Local differences in the sequences are identified in all syntenic and rearranged regions. SNPs and small indels are parsed directly from the local alignments, whereas more complex sequence variations (e.g., like large indels and CNVs) are identified in the overlaps and gaps between consecutive local alignments.” “all non-aligned regions in between syntenic and rearranged regions are reported for completeness.”

Terminology: “For simplicity, unless specified otherwise, we refer to transpositions and translocations together as ‘translocations’.” (Transposition = intra-chromosomal, translocation = inter-chromosomal is the conventional reading and consistent with Fig 2’s separate “inter-chromosomal rearrangements” step, but the paper does not spell out that intra/inter distinction in the retrieved text — [summarizer-inferred].)

6. Key claims / findings

“SyRI … is the first tool to identify all structural and sequence differences between two chromosome-level genome assemblies.”
“SyRI annotates the coordinates of rearranged regions (i.e., breakpoints on both sides of a rearrangement in both genomes).”
Experimental validation: “Out of 117 translocations, 108 (92.3%) could be confirmed by at least one test.” (These 117 translocations were “larger than 1 kb” — see §12 Q4.)
Runtime/memory: “For the two human genomes SyRI took ~ 10 min, while memory usage was less than 1 GB for each.”
Benchmark: compared on 100 simulated genomes against six tools — AsmVar, Smartie-sv, Assemblytics, Sniffles, Picky, LUMPY (Fig 3). Indel sensitivity/precision evaluated at two error limits, 5 bp and 100 bp. “Not Applicable” = tool not designed for that difference type.
Requirement (explicit): “SyRI requires whole-genome alignments from chromosome-level assemblies as input.”
Incomplete-assembly path: “If one or both of the assemblies is/are incomplete, pseudo-chromosomes can be generated using homology between the assemblies themselves or using homology to a chromosome-level reference sequence.” And: prediction of genomic rearrangements “is nearly complete even if one of the genomes is not on chromosome-level, but has assembly contiguity of N50 > 500 kb.”

7. Load-bearing statements — VERBATIM (CC BY, reproduction permitted)

“SyRI expects whole-genome alignments (WGA) as input” — Methods/Fig 2.
“The syntenic path represents the longest set of non-rearranged regions between two genomes.” — Fig 2 legend.
“Since all non-syntenic regions are rearranged by definition, identifying syntenic regions identifies rearranged regions at the same time.” — Results.
“SyRI requires whole-genome alignments from chromosome-level assemblies as input.” — Discussion/limitations.
“SNPs and small indels are parsed directly from the local alignments, whereas more complex sequence variations (e.g., like large indels and CNVs) are identified in the overlaps and gaps between consecutive local alignments.” — Fig 2 legend / Step 3.

8. Stated scope, assumptions, limitations (source’s own)

Pairwise only (two genomes at a time); designed for chromosome-level assemblies.
Requires a WGA as input; the tool consumes alignment coordinates, not raw reads.
For incomplete input, the user must first build pseudo-chromosomes (via mutual homology or homology to a chromosome-level reference); SyRI does not scaffold for you.
Rearrangement calling degrades gracefully: “nearly complete” when one genome is only at N50 > 500 kb, i.e. below chromosome-level but still highly contiguous.

9. Failure modes / invalidity patterns

Non-chromosome-level input without pseudo-chromosomes. SyRI’s syntenic-path spine runs between the two ends of a chromosome; fragmented/contig input breaks this. The stated remedy is pseudo-chromosome construction. (Paper frames this as a requirement, not a warned-against silent failure.)
Filtering of ambiguous rearrangement candidates: “Low scoring candidates and those that are overlapping with syntenic or inverted regions are filtered out.” → real rearrangements overlapping syntenic/inverted blocks can be dropped; conversely this is the paper’s stated guard against spurious calls.
Alignment-quality dependence [summarizer-inferred]: because SyRI consumes a WGA, its output inherits the underlying aligner’s errors; the paper does not, in the retrieved text, give an explicit caveat quantifying false-positive rates per rearrangement class.
No explicit false-positive-inversion / over-calling caveat in the retrieved main text (see §12 Q6, ABSENT).

10. What the source does NOT address (in retrieved text)

An enumerated table of output annotation abbreviations (SYN/TRANS/INVTR/DUP/INVDP/CPG/ CPL/NOTAL/TDM, etc.) — NOT present in the paper prose or figure legends retrieved (§12 Q2).
Output filenames syri.out / syri.vcf and their column schema — NOT named in the retrieved main text (§12 Q7); these are software-documentation artifacts.
A minimum size threshold for calling inversions or duplications (§12 Q4).
Direct SAM/BAM input handling / a -F format flag (§12 Q1) — the guidance’s -F B = BAM is a software-CLI detail, not stated in the paper.
Exact nucmer -c/-b/-l values (deferred to Additional file 2: Table S3, not retrieved).

11. Open questions / ambiguities

Precise intra- vs inter-chromosomal definitions for “transposition” vs “translocation” (paper collapses them under “translocations” for simplicity).
Whether the abbreviation set + syri.out/syri.vcf schema the guidance expects come from the SyRI software docs/README rather than this paper (they are not in the article).

12. Guidance answers

Q1 — Input & aligners / flags. Input = whole-genome alignments (WGA). Aligner: MUMmer3 nucmer run with --maxmatch plus -c, -b, -l (exact values in Additional file 2: Table S3, not retrieved); coords produced via delta-filter/show-coords. minimap2 “can be used as well.” The paper’s retrieved text does NOT document a BAM/SAM path or the -c/-r/-q/-F/--prefix/-k CLI flags — those are SyRI-software CLI details, not paper content (so -F B = BAM cannot be confirmed from this source; ABSENT here).

Q2 — Structural-variant abbreviation set (must-extract). The paper does NOT enumerate the abbreviation set. Retrieved prose/figures use full words (“inversions,” “transpositions,” “translocations,” “duplications,” “syntenic,” “highly divergent regions,” “non-aligned regions”) plus a few abbreviations: INV, DUP, HDR, SNP, WGA, CNV. The guidance’s expected set — SYN, INV, TRANS, INVTR, DUP, INVDP, INS, DEL, HDR, CPG, CPL, NOTAL, TDM — is ABSENT from this paper; it is a SyRI-output/documentation convention that must be sourced from the tool’s docs/README, not from Goel 2019. (Faithful negative: do not attribute that table to this paper.)

Q3 — Syntenic-path algorithm. SyRI finds “the highest scoring syntenic path” — “the longest set of non-rearranged regions between two genomes” — by dynamic programming from one chromosome end to the other; then, because “all non-syntenic regions are rearranged by definition,” everything off-path is a rearrangement, classified into inversions, transpositions, duplications, and inter-chromosomal rearrangements (Step 2). See §5 quotes.

Q4 — Minimum size for inversion/translocation/duplication. ABSENT. The paper states no minimum-size threshold for inversions or duplications. The only size figure is descriptive of a validation set: “Out of 117 translocations” that “were larger than 1 kb” — this is the size of the experimentally-tested translocations, NOT a calling threshold. → Any skill’s ”≥ 5 kb (biological significance)” inversion cutoff is NOT supported by this paper; treat it as convention/software-default, not a Goel-2019 claim.

Q5 — Chromosome-level requirement. YES, stated: “SyRI requires whole-genome alignments from chromosome-level assemblies as input.” Fragmented/contig-level input: build pseudo-chromosomes first (via mutual homology or homology to a chromosome-level reference). Rearrangement prediction is “nearly complete even if one of the genomes is not on chromosome-level, but has assembly contiguity of N50 > 500 kb.” The retrieved text does NOT explicitly require equal chromosome counts or identical chromosome IDs — ABSENT (not stated either way).

Q6 — False-positive / inversion over-calling caveat. No explicit caveat on inversion over-calling or strand/assembly artifacts in the retrieved main text (ABSENT). The nearest stated control is the Step-2 filter: “Low scoring candidates and those that are overlapping with syntenic or inverted regions are filtered out.” Validation reports 92.3% (108/117) translocations confirmed — evidence of low FP for translocations, but nothing inversion-specific.

Q7 — Output files & column semantics. The retrieved paper text does NOT name syri.out or syri.vcf nor describe their columns (ABSENT). It states only that SyRI “annotates the coordinates of rearranged regions (i.e., breakpoints on both sides of a rearrangement in both genomes).” Output-file/format specifics must come from the SyRI software docs, not this article.

design-inference: none in this file (kept separate per convention).