Omega-seq: ultra-low-background RNA sequencing with faithful molecular counting and precise transcript-end capture

Key results

1. 01

   ShownOmega-seq showed less UMI inflation than FLASH-seq-UMI on diluted human RNA by two readouts: a new rarefaction metric (logQ-slope, the log-log slope of unique UMIs per UMI-bearing read against depth, near -1 when UMIs are fixed before PCR) and ERCC spike-in counts.

   Human total RNA, 10/1/0.1 pg: logQ-slope about -0.3 for FLASH-seq-UMI (published oligo concentrations) vs approaching -0.8 for Omega-seq (n = 8 vs n = 8, p < 2.5 × 10^-7). ERCC UMI/input ratios exceeded 1 for many FLASH-seq-UMI transcripts (up to 172), stayed below 1 for Omega-seq. Primer-to-TSO ratios differ roughly 12-fold (0.27 vs 3.3); authors note this alone could explain the slope gap. · Rarefaction and ERCC analyses demonstrate high…; Illumina sequencing; Analysis; Figure legends

2. 02

   ShownOmega-seq oligonucleotides produce almost no non-specific amplification, so low-input preamplified cDNA is nearly all mappable without cleanup. At low input, FLASH-seq-UMI product is dominated by short artifacts that its bead cleanup has to remove.

   Drosophila total RNA, matched buffer and oligo load: no-template controls of Smart-seq2, Smart-seq3, FLASH-seq-LA and FLASH-seq-UMI often reached cycle thresholds near low-input samples, while Omega-seq separated 10 fg from controls. Nanopore of unpurified product: >99% of Omega-seq reads mapped at every non-zero input vs 65%, 16% and 2% for FLASH-seq-UMI at 10, 1 and 0.1 pg. · Omega-seq suppresses non-specific amplification…; Nanopore characterization of preamplified cDNA; qPCR; Nanopore sequencing

3. 03

   SuggestedIn individual Drosophila neuroblasts, Omega-seq logQ-slopes were tighter and steeper than in published Smart-seq3 data run with high primer or bead cleanup. This points to the combined uracil and polymerase barriers curbing PCR-stage UMI creation more than excess primer or bead removal alone.

   Omega-seq neuroblasts (n = 309 cells, one plate) vs Smart-seq3 HEK293 at 1 µM forward primer (n = 48; p < 10^-80) and bead-cleaned at 0.1 µM (n = 48; p < 10^-18). Companion-study clone-size estimate ŵtso ≈ 0.05 for Omega-seq vs ≈ 0.4 for FLASH-seq-UMI and Smart-seq3 at 0.1 µM; per-library ranking agreed with logQ-slope (Spearman ρ = 0.90). · Rarefaction and ERCC analyses demonstrate high…; Analysis

4. 04

   ShownAn oligo-dT primer with an internal sequencing handle (Omega-dT) avoids sequencing through synthetic poly-T. Its reads are mappable and stranded and pinpoint transcription end sites (TES).

   dT reads from a conventional oligo-dT with s5+ME appended mapped poorly because of poly-T dephasing; Omega-dT restored dT-read mappability (slightly below TSO and internal reads), with reads clustering at annotated TES and the upstream AATAAA polyadenylation signal. A-rich internal-priming reads are filtered; lower Tn5 concentration enriched start- and end-site reads. · Omega-dT enables stranded capture of…; Figure legends

5. 05

   SuggestedPooling stranded end and internal reads across Drosophila neuroblasts and neurons allowed reference-free gene-model reconstruction, which flagged candidate unannotated transcripts and 3′ UTR extensions.

   Splice-graph pipeline using junctions with ≥10 spliced reads plus sequence-context-filtered TSO/dT ends. Reported examples: 3′ UTR extensions differing between neuroblasts and neurons, opposite-strand overlapping transcripts, neuroblast-only candidate enhancer RNAs, unannotated multi-exon genes and intronic antisense transcripts; no genome-wide counts or orthogonal validation given. · Stranded, end-delineated reads enable de novo…; Analysis