Skip to content

Sydney Blattman — GIFT: Genotyping In Fixed Transcriptomes

Speaker: Sydney B. Blattman — postdoc, Dana Pe'er Lab, Memorial Sloan Kettering Cancer Center (PhD, Columbia); collaboration with the Caleb LaReau Lab + MSK Single-Cell Analytics Innovation Lab and 10x Genomics. Shout-outs: Nabi (Nabih Maslah), Austin (Austin A. Varela) — both preprint co-authors. Talk title (SCGD26 program): "Scalable genotyping in fixed transcriptomes resolves clonal heterogeneity" Slot: SCGD26, Jun 12, 2026 Method: GIFT (Genotyping In Fixed Transcriptomes) / GIFT-seq Recording: SCGD26 livestream; Q&A answered live by Caleb LaReau Preprint: Blattman et al., "Scalable genotyping in fixed transcriptomes resolves clonal heterogeneity via single-cell sequencing," bioRxiv 2026 — doi:10.64898/2026.04.11.717967 · PubMed 41993258 Protocol: GIFT-seq Gapfill and Library Preparation Protocol (protocols.io) Status: Captured from session transcript — preprint + protocols.io live (confirmed)

Name note: the SCGD26 program and the bioRxiv author list both spell the speaker Blattman (not "Blackman" as auto-transcribed); affiliation is MSKCC (Pe'er / Chaligné / LaReau labs).

Thesis

Read somatic genotype and phenotype in the same single cell, at scale, including from FFPE/archival tissue. Somatic mutations drive tumorigenesis, metastasis, drug resistance (and possibly autoimmunity); many have undetected phenotypes. GIFT pairs mutation calls with gene-expression for clonal/lineage reconstruction.

How it works

Built on 10x Flex probe-based chemistry (not reverse transcription). Standard Flex ligates two single-stranded DNA probes hybridized to an mRNA. GIFT introduces a gap (up to ~10 bases) between the two probes, adds a polymerase to fill the gap over the genotyping site, then ligates → the variant is read out as a single-cell-barcoded molecule alongside normal Flex gene-expression probes.

Addresses the three limits of GOT (Genotyping Of Transcriptomes, 2019): low variant throughput, restriction to transcript-end mutations, and FFPE-incompatibility.

Key results

  • Accuracy: cell-line mixing (K562 / SET-2 / etc.) → >99% genotyping accuracy; correctly calls heterozygous cells (needs ≥1 WT + ≥1 mutant count — a real test of capture rate).
  • Scale: >600 probe pairs → median ~164 genotyped sites per cell (orders of magnitude over GOT); essentially no transcript-position bias.
  • Best-of-both vs. "dual probes" / GOT-multi: allele-specific hybridization probes give degraded accuracy with a long failure tail; GIFT's polymerization step keeps ~99% accuracy.
  • FFPE: a de-crosslinking step raised gene-expression yield ~10× (from ~250 counts/cell) and genotyping-probe capture ~10×. Applied to an FFPE glioblastoma sample: three EGFR mutations resolved into clones (one EGFR co-occurs with an AXL mutation = same clone; the other two are mutually exclusive = independent clones) → coarse lineage tree beyond bulk VAF.
  • Cohort scale: 35 myeloproliferative-neoplasm patients, ~700k cells, 66 loci. JAK2 genotype tracks HSC + monocyte clusters; dosage response (WT → het → homozygous via LOH) with sequentially increasing interferon-γ response. An MPN→AML case reconstructed a CALR/TP53/ZRSR2/EZH2 lineage tree.

Q&A highlights

  • RNA vs DNA genotyping (Caleb LaReau): sensitivity scales with expression; for pre-malignant rare-mutation cases, DNA may be fundamentally better, but many target genes are well-expressed and the win is multi-omics + the ability to load millions of cells (Flex v…) to recover rare cells.
  • Beyond SNVs: indels/large variants handled by staggering multiple probes (ratio read-out) rather than gap-fill (see preprint Fig 3, EGFR example).
  • Spatial / Visium HD: the preprint confirms GIFT is Visium HD–compatible — adapted to a spatially-restricted three-cell-line pattern with 98.8–99.7% accuracy, though spatial GE/GIFT complexity was lower than dissociated single cells, so most applications used the dissociated workflow.

Connections to the corpus

  • Single-cell genotype↔phenotype + lineage, FFPE-compatible — directly relevant to AACR single-cell-spatial-omics and clinical clonal-evolution work.
  • Flex/probe-chemistry sibling to Alex — VIP Perturb-seq (also Flex-based) and Caleb LaReau's perf-seq (RNA-FISH enrichment).

Sources

All header attributions confirmed against the SCGD26 program and the bioRxiv preprint; no open items remain.