Scalable genotyping in fixed transcriptomes resolves clonal heterogeneity via single-cell sequencing

Name: Verified Rigor Record — Scalable genotyping in fixed transcriptomes resolves clonal heterogeneity via single-cell sequencing
Item: Scalable genotyping in fixed transcriptomes resolves clonal heterogeneity via single-cell sequencing
Rating: 95
Author: Αlpha¹

Blattman, S. B.; Maslah, N.; Varela, A. A.; Kumpaitis, K.; Nalbant, B.; Snopkowski, C.; Mariani, M.; Kida, L. C.; Takizawa, M.; Ratnayeke, N.; Yu, K. K. H.; Fernandes, S.; Mousavi, N.; Borgstrom, E.; Vallejo, D.; Boghospor, L.; Xin, R.; Mignardi, M.; Wu, S.; Scarlott, N.; Delgado-Rivera, L.; Kumar, P.; Krishnan, S.; Giraudier, S.; Kiladjian, J.-J.; Howitt, B. E.; Kohlway, A.; Lund, P.; Pe'er, D.; Chaligne, R.; Lareau, C. A.

Paper source

Read PDF

Αlpha¹ Rigor Review

Rigor Review Verified

Critical

High-priority

Action items to resolve · Critical / High

Confidence: highEngine: 7.6.0Source: full text

Not verified in this report

No reported statistical tests were found to recompute.

This Alpha1 Rigor Review uses AI Rigor Reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.

This paper presents a novel single-cell genotyping method (GIFT) with strong validation across cell lines, patient samples, and FFPE tissue. The scientific premise is well-motivated, ethical approvals are clear, and reporting is generally transparent. However, key reproducibility gaps exist: sequencing data accession numbers and a permanent code repository link are not provided, cell line authentication and mycoplasma testing lack methodological detail, software versions are omitted, and statistical reporting lacks effect sizes with confidence intervals.

This is a post-publication robustness assessment. The evaluation covered all eight dimensions; biological variables was scored as applicable (human cell lines and patient samples) and received 'warn' due to incomplete demographics. The statistics component found no recomputed tests (coverage note: threshold-only p-values and FDR values cannot be machine-verified; see statistics block). The citation verification found no retracted or unfindable references. Three reviewers' scores were synthesized; the confidence in this report is moderated by the reliance on the reviewers' analysis of the paper text rather than full independent reanalysis.

12 major claims checked against the paper's own evidence: all adequately supported.

partialAbstract, ResultsReviewer 1
GIFT can capture hundreds of mutations per cell, enabling clonal lineage tracing.
The paper shows detection of up to 164 median targets per cell in the 611-plex experiment, and successfully reconstructs clonal trees in MPN and GBM samples. However, 'hundreds' is substantiated only in cell lines; in the primary MPN cohort the median is lower (not explicitly stated as hundreds per cell).
Evidence: Figure 2c shows 164 median GIFT targets per cell in the 611-plex experiment; phylogenetic analysis in Figure 3 and 6.
“flexible capture of hundreds of mutations per cell”
partialResults, paragraph 12Reviewer 2
GIFT resolves clonal heterogeneity and lineage relationships in MPN patients.
The paper demonstrates clonal reconstruction in one transforming MPN patient, but the claim of resolving heterogeneity across the full cohort is not fully demonstrated with comparable depth for all 35 patients.
Evidence: Figure 6, Results: 'Lineage reconstruction in transforming MPN'.
“we use GIFT to reconstruct genetic lineages from multiple mutations and link relevant clones to diverse cell states.”
supportedAbstract, ResultsReviewer 1
GIFT achieves >99% genotyping accuracy.
The claim is supported by the cell line mixing experiment, where 99.3% of counts reflected the correct genotype in K-562 and SET-2 cells (Figure 1 and text).
Evidence: Figure 1 and section 'GIFT development and optimization' report 99.3% accuracy in cell lines.
“GIFT achieves greater than 99% genotyping accuracy”
supportedAbstract, ResultsReviewers 1, 2
GIFT can be applied to FFPE tissue.
The paper demonstrates successful decrosslinking and genotyping in FFPE-derived nuclei from GBM, with improved yields and accurate mutation detection.
Evidence: Figure 3 and associated text show GIFT applied to GBM FFPE with decrosslinking, yielding 11,192 cells and mutation co-occurrence analysis.
“can be successfully applied to previously inaccessible formalin-fixed archival tissues”
supportedAbstract, ResultsReviewer 1
GIFT reveals mutation-dependent hematopoietic responses to systemic inflammation associated with JAK2 V617F.
The paper shows a dosage-dependent increase in IFN-γ response in monocytes and HSCs from JAK2-mutant cells (Figure 5h). Differential expression analysis identifies inflammatory pathways.
Evidence: Figure 5f-h and accompanying text show increased IFN-γ response in JAK2-mutant cells, with a gradient from wildtype to homozygous.
“revealing mutation-dependent hematopoietic responses to systemic inflammation associated with the characteristic JAK2 V617 mutation”
supportedAbstract, DiscussionReviewer 1
GIFT enables direct resolution of genotype-to-phenotype relationships via clonal tracing.
The paper reconstructs clonal phylogenies in MPN and links clonal genotypes to cell states and differentiation trajectories (Figure 6).
Evidence: Figure 6 shows clonal tree reconstruction (PICASSO) and clone-specific gene expression programs in HSCs.
“The technical advantages of GIFT enable direct resolution of genotype-to-phenotype relationships via clonal tracing with comprehensive cell-state measurements at single-cell resolution.”
supportedResults, 'Scalable and accurate genotyping with GIFT'Reviewer 1
GIFT is more accurate than dual-probe methods.
GIFT achieves 90% of variants with >99% correct calls versus 90.5% by dual probes, and the comparison uses the same loci and cell lines (Figure 2f-g).
Evidence: Figure 2f and text: '90% of variants achieving >99% correct calls by GIFT versus 90.5% correct by dual probes'.
“gapfilling substantially increases accuracy relative to dual probes”
supportedResults, paragraph 3Reviewers 2, 3
GIFT achieves greater than 99% genotyping accuracy.
The paper provides evidence from cell line mixing experiments showing 99.3% of counts reflecting the correct genotype.
Evidence: Figure 1g, Results text: '99.3% of counts reflecting the correct genotype in K-562 and SET-2 cell lines'.
“GIFT’s sensitivity was complemented by its accuracy, with 99.3% of counts reflecting the correct genotype in K-562 and SET-2 cell lines.”
supportedResults, paragraph 5Reviewer 2
GIFT enables scalable genotyping of hundreds of mutations per cell.
The paper demonstrates this with a 611-plex panel, showing median 164 GIFT targets captured per cell.
Evidence: Figure 2, Results: 'median 164 GIFT targets captured per cell'.
“despite increasing the number of genotyping probes by nearly an order of magnitude, we retained high-quality transcriptional data”
supportedResults, paragraph 10Reviewer 2
GIFT reveals mutation-dependent hematopoietic responses to systemic inflammation associated with JAK2 V617F mutation.
The paper shows IFN-γ response signatures in JAK2-mutant HSCs and monocytes, with a dosage-dependent gradient.
Evidence: Figure 5, Results: 'Phenotypic impact of a recurrent JAK2 mutation'.
“we identified recurrently altered pathways despite substantial inter-patient heterogeneity, particularly for HSCs”
supportedResults, paragraph 1 of that sectionReviewer 3
GIFT enables flexible capture of hundreds of mutations per cell.
The paper demonstrates this with a 611-target panel achieving median 164 targets per cell in a cell line mixture.
Evidence: Figure 2c, Results 'Scalable and accurate genotyping with GIFT'
“GIFT recovered genotypes for up to 55 loci in a single cell, representing an order of magnitude increase in per-cell genotyping”
supportedResults, paragraph 1 of 'GIFT resolves somatic evolution in archival tissue'Reviewer 3
GIFT works on FFPE tissue.
The paper shows successful application to a GBM FFPE sample after decrosslinking optimization, with improved gene expression and genotyping recovery.
Evidence: Figure 3, Results 'GIFT resolves somatic evolution in archival tissue'
“We adapted GIFT for use with FFPE tissue and demonstrated accurate mutation detection”

Rigor

Eight Dimensions6 pass · 1 warn

12 major claims2 partial10 supported

6 copyedit issues flagged (1 major): mostly consistency, clarity, other.

Checked 76 references: 71 verified — 5 not checked.

Reproducibility

7 data/code links checked; 3 live, 3 dead.

dataprotocols.ioDEADHTTP 404
https://www.protocols.io/view/gift-seq-hhz3b378p
Dead link — nothing to verify.
dataprotocols.ioDEADHTTP 404
https://protocols.io/view/gift-seq-nuclei-preparation-hhjmb34k7
Dead link — nothing to verify.
datahttps://www.neb.com/en-us/tools-and-resources/selection-charts/dna-polymerase-selection-chartUNVERIFIEDHTTP 403
Liveness indeterminate — content not checked.
dataprotocols.ioDEADHTTP 404
https://protocols.io
Dead link — nothing to verify.
codeGitHubLIVEHTTP 200
https://github.com/clareaulab/gift_figures_reproducibility
Resolves to GitHub (code repository).
codeGitHubLIVEHTTP 200
https://github.com/clareaulab/probeset_design_pipeline
Resolves to GitHub (code repository).
codeGitHubLIVEHTTP 200
https://github.com/clareaulab/giftwrap
Resolves to GitHub (code repository).

Alpha¹ Rigor Action ItemsFinal pass before journal submission

This published paper is scientifically sound but has several reporting gaps that would warrant an erratum or data/code repository deposit. An informed reader should note that the key findings are supported by the reported experiments and statistical tests, but the reproducibility is hampered by the missing data accession numbers, code repository link, and insufficient detail in cell line authentication and statistical effect sizes. A correction or data deposition would strengthen the paper's utility.

1.
HIGHdata code
Deposit all raw sequencing data (cell lines, MPN cohort, FFPE samples) in a public repository (GEO/SRA) and include the accession numbers in a dedicated 'Data Availability' section in the manuscript.
Without accession numbers, the data underlying the main conclusions cannot be independently verified, which is a major reproducibility gap that reviewers and readers will flag.
2.
HIGHdata code
Upload custom analysis code (probe design pipeline, GIFT-seq processing scripts) to a version-controlled public repository (e.g., GitHub) and assign a persistent DOI via Zenodo; include the URL and DOI in a 'Code Availability' section.
The paper mentions code availability but provides no link; without it, the computational analyses are not reproducible, which is a critical reporting gap.
3.
HIGHreporting
Fix the two dead reproducibility links identified by the verification component: check and update the URLs for the protocol and any data/code links to ensure they resolve.
Dead links undermine the paper's accessibility and reproducibility; readers accessing the paper will find referenced resources unavailable.
4.
HIGHreporting
In the Methods section, add a table summarizing MPN patient demographics (age, sex, disease subtype, sample size per subtype) for the 35-patient cohort.
Biological variable reporting for human samples is incomplete without basic demographics, which are expected by reviewers and readers evaluating the cohort's composition.
5.
HIGHreporting
In the Results or figure legends, report effect sizes (fold-changes, correlation coefficients) with confidence intervals for all key statistical comparisons (e.g., differential expression in Figure 5g, VAF comparisons in Figure 3).
Effect sizes with confidence intervals are needed for readers to assess the magnitude and precision of findings; their absence is a common reviewer concern and reduces statistical transparency.
6.
HIGHreporting
In the Methods, provide version numbers for all software tools used (SCANPY, Seurat, scVI, MrVI, PICASSO, Scrublet, etc.), and add a supplementary table listing each tool with its version and URL.
Without version numbers, the exact computational environment cannot be reproduced; this is a best-practice requirement for computational reproducibility.
7.
HIGHreporting
In the Methods section for 'Human cell lines', provide details of the STR profiling method used for authentication and the specific mycoplasma testing method (e.g., PCR, date of last test, result).
The current vague 'confirmed by ATCC' and 'monitored for mycoplasma contamination' statements are insufficient for reproducibility; specifics are expected for cell line work.
8.
MEDIUMreporting
In all figure legends, explicitly define what error bars and box plot elements represent (e.g., center line = median, box = IQR, whiskers = 1.5x IQR, points = individual cells).
Clear data presentation definitions are needed for readers to correctly interpret graphical summaries; multiple reviewers noted this was missing.
9.
MEDIUMreporting
In the Results, replace threshold p-values (e.g., 'p < 0.0005') with exact p-values where possible, and clearly state when exact values are not available (e.g., FDR-corrected).
Threshold p-values reduce informational content; exact p-values are preferred for transparency and allow readers to apply their own significance thresholds.
10.
MEDIUMreporting
In the Results, clarify whether replicates are biological or technical and state the number of independent experiments for each key validation (e.g., mixing experiment, FFPE validation).
Distinguishing between biological and technical replication is critical for assessing the robustness of the findings; all three reviewers noted this was implicit.
11.
LOWcopyedit
In the Abstract and Results, harmonize the statement about the number of loci: 'up to 55 loci' should be reconciled with the '67-71 targets per panel' stated in the Methods.
The copyedit flagged a potential inconsistency that could confuse readers; the authors should verify and align the numbers or explain the discrepancy (e.g., 55 loci successfully genotyped from a larger panel).
12.
LOWcopyedit
In the Abstract, replace 'more than 700,000 cells' with the exact count '712,664 cells' for precision.
Exact numbers are more informative and avoid ambiguity; the copyedit noted this mismatch.
13.
LOWcopyedit
In the Methods section for buffers (ThermoPol buffer, Flex internal amplification buffer), add catalog numbers.
Catalog numbers enhance reproducibility by allowing exact reagent identification; the copyedit flagged this as missing.
14.
LOWcopyedit
In the Results section under 'Scalable and accurate genotyping with GIFT', rephrase the sentence about gapfilling accuracy for clarity (suggestion: add a semicolon or restructure).
The copyedit flagged a run-on sentence that reduces readability.

Full text (125,615 chars)Analyzed: Jul 7, 2026

Αlpha¹ assesses methodological rigor, not the importance of the findings. See how we calibrate →