Gene conversion is a key driver of diversity hotspots in M. tuberculosis antigens and virulence-associated loci
Marin, M. G., Quinones-Olvera, N., Jin, H., Harris, M. A., Jeffrey, B. M., Rosenthal, A., Murphy, K. C., Sassetti, C., et al.
genomics
Quick verdict
Claims paralog gene conversion drives Mtb diversity hotspots incl. antigen PPE18; well-supported long-read evidence, some sampling and sensitivity caveats.
Summary
Using 151 complete M. tuberculosis genome assemblies built from matched long- and short-read data, the authors quantify genome-wide nucleotide diversity and show that 36 of 37 diversity hotspots (12-83x the genomic median) fall inside paralogous regions enriched for PE/PPE, ESX, and REP13E12 genes. A Gubbins-based phylogenetic workflow detects 324 gene conversion events (median tract 101 bp), 91% within paralogous regions and 74% mappable to donor paralogs; five high-burden paralog networks account for 55% of events, which occur on both recent and ancestral branches with lineage-specific rate variation. Assembly-error QC (manual curation plus NucFlag) and independent validation - short-read screening of 937 isolates followed by PacBio HiFi resequencing of 22 - support the calls while showing short-read sensitivity is only ~14%. PR-encoded T-cell antigens carry roughly 4x higher mutational burden than unique antigens (median 4.1 vs 1.0 per 100 aa), and 25 of 295 in-PR events alter validated epitopes. In vaccine antigen PPE18, 44 of 61 non-synonymous mutations hit epitopes and HLA-II binding changes are predicted, though the epitope-concentration trend is non-significant (p=0.055). Limitations include underdetection between near-identical paralogs, single-reference bias, sparse sampling of some lineages, and immunology resting on in silico prediction.
Review complete
Assessment at a glance
4
risk flags identified
● high 0
● medium 2
● low 2
5 strengths · 5 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Recurrent inter-paralog gene conversion, not point mutation, generates Mtb diversity hotspots: 36 of 37 hotspots (12-83x genomic median diversity) lie in paralogous regions.
- 324 gene conversion events are detected across the Mtb phylogeny, concentrated in PE/PPE, ESX, and REP13E12 loci, with five paralog networks carrying 55% of events.
- T-cell antigens encoded in paralogous regions bear about 4-fold higher mutational burden than unique-region antigens, and 25 of 295 in-PR gene conversion events alter validated epitopes (5.0-fold enrichment).
- The vaccine antigen PPE18 carries 44 of 61 non-synonymous mutations within validated epitopes, and five of seven detected gene conversion events are predicted to alter HLA-II binding.
Strengths
Resolves the portion of the Mtb genome all prior short-read analyses excluded
151 complete hybrid assemblies (lineages 1-6 and 8) expose 37 diversity hotspots (2.3-15.5 subst/kb vs 0.2 median), 36 of which sit in paralogous regions - the PE/PPE, ESX, and REP13E12 loci routinely masked in short-read data.
Assembly-artifact controls are unusually thorough
Manual inspection of all 37 hotspots across all genomes plus reference-free NucFlag screening: 0 of 24,463 hotspot SNPs overlapped flagged regions (vs 29/220,430 genome-wide, 0.013%), directly countering the artifact objection (Dataset S15).
Independent experimental validation with honest sensitivity accounting
Convergent, multi-layer evidence rather than a single-tool result
Statistical transparency including the authors' own null results
Weaknesses
Event detection is threshold-conditioned with no sensitivity analysis
Lineage rate comparisons rest on very unequal sampling
Internal inconsistency in the mutational-spectrum result
Immunological claims rest entirely on in silico prediction
Minor numeric inconsistencies and a whiff of circularity
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Scope & Claims
This bioRxiv preprint attacks a long-standing blind spot in M. tuberculosis genomics: the ~10% of the genome - PE/PPE, ESX, and repeat loci - that short-read pipelines cannot resolve. Using 151 complete hybrid assemblies spanning lineages 1-6 and 8, the authors claim that (1) 36 of 37 extreme diversity hotspots (Modified Z > 10; 2.3-15.5 subst/kb against a 0.2/kb median, i.e., 12-83x) sit inside paralogous regions; (2) recurrent intrachromosomal gene conversion generates them, evidenced by 324 phylogenetically placed events (median tract 101 bp, 6 SNPs), 91% within paralogous regions and 74% mappable to donor paralogs; (3) PR-encoded T-cell antigens carry ~4x the mutational burden of unique-region antigens (median 4.1 vs 1.0 per 100 aa, Table 4); and (4) the vaccine antigen PPE18 (a component of M72/AS01E and BNT164) accumulates epitope mutations with predicted HLA-II binding consequences. Claims are appropriately scoped to genomic association, not immunological causation.
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 5 strengths & 5 weaknesses, analyzed in detail
- 4 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
What is gene conversion in M. tuberculosis and why does it matter?
Gene conversion is a non-reciprocal recombination process in which a short DNA tract is copied from one paralogous gene to another within the same genome. In M. tuberculosis - long viewed as evolving almost purely by point mutation - this paper shows it recurrently reshapes paralogous PE/PPE, ESX, and repeat loci, generating diversity hotspots up to 83 times the genomic background, including in antigens relevant to TB vaccines.
Is PPE18, a component of the M72/AS01E and BNT164 vaccine candidates, variable between Mtb strains?
Yes. Across 151 complete genomes, PPE18 accumulated 61 non-synonymous mutations, 44 of which fall inside experimentally validated T-cell epitopes, driven by seven detected gene conversion events. Computational HLA-II binding predictions suggest several events alter peptide presentation, though the immunological consequences remain untested and the epitope-concentration trend itself was not statistically significant (p=0.055).
Why did previous Mtb genomic studies miss these diversity hotspots?
Roughly 10% of the Mtb genome is repetitive or paralogous and cannot be reliably analyzed with short reads. In this study's validation, short-read data recovered only 86 of 199 SNPs resolved by long-read assemblies (event recall 14.3%), whereas complete long-read-based assemblies resolved the full events - explaining why prior short-read-based analyses concluded Mtb antigens were nearly invariant.
How did the authors rule out assembly errors as the cause of the hotspots?
Two complementary controls: manual inspection of long-read and assembly-to-reference alignments at all 37 hotspots across all 151 genomes, and reference-free misassembly screening with NucFlag. None of the 24,463 SNPs in hotspots overlapped a NucFlag-flagged region, while only 29 of 220,430 SNPs genome-wide (0.013%) did.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Text extraction corrupted several p-value exponents (e.g., p_adj = 9.6 x 10^-, hypergeometric p = 9.7 x 10^11) and part of the Figure 1 legend; exact significance levels could not be verified from this text.
- Figures, supplementary methods/results, and most supplementary tables were not analyzable; judgments rest on the main text and Tables 1-4.
- This is a bioRxiv preprint (10.64898/2026.02.26.708061) and has not completed peer review.
- The 18,371 vs 18,325 peptide-count discrepancy between Results and the Figure 5 legend could be an extraction artifact but should be checked against the published PDF.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.