Context-Aware Transcript Quantification from Long Read RNA-Seq data with Bambu

Abstract

Most approaches to transcript quantification rely on fixed reference annotations. However, the transcriptome is dynamic, and depending on the context, such static annotations contain inactive isoforms for some genes while they are incomplete for others. To address this, we have developed Bambu, a method that performs machine-learning based transcript discovery to enable quantification specific to the context of interest using long-read RNA-Seq data. To identify novel transcripts, Bambu employs a precision-focused threshold referred to as the novel discovery rate (NDR), which replaces arbitrary per-sample thresholds with a single interpretable parameter. Bambu retains the full-length and unique read counts, enabling accurate quantification in presence of inactive isoforms. Compared to existing methods for transcript discovery, Bambu achieves greater precision without sacrificing sensitivity. We show that context-aware annotations improve abundance estimates for both novel and known transcripts. We apply Bambu to human embryonic stem cells to quantify isoforms from repetitive HERVH-LTR7 retrotransposons, demonstrating the ability to estimate transcript expression specific to the context of interest.

Merging short and stranded long reads improves transcript assembly

“…First, we compared TASSEL to other hybrid and standalone long-read transcript assembly programs for transcript assembly of ERCC spike-in standards (Fig 5B). FLAIR [53], Bambu [65] and String-Tie Mix [29] failed to assemble ∼40 of the 92 ERCC transcripts, while only 13 transcripts were not assembled via TASSEL. FLAIR and StringTie Mix also assembled many incomplete ERCC transcripts where the assembled transcripts were shorter than the actual molecular standards.…”

“…Informed by the strengths and caveats of each of the two platforms, we developed a hybrid pipeline, TASSEL, to merge the short-read transcriptome with the stranded long-read transcriptome. We tested TASSEL against commonly used long-read transcript assembly methods such as FLAIR [53], Bambu [65] and StringTie Mix [29]. TASSEL had the lowest false negative rate of transcript assembly and unlike other methods, TASSEL led to assembly of complete transcripts on the correct strand.…”

“…First, we compared TASSEL to other hybrid and standalone long-read transcript assembly programs for transcript assembly of ERCC spike-in standards (Fig 5B ). FLAIR [53], Bambu [65] and String-Tie Mix [29] failed to assemble ~40 of the 92 ERCC transcripts, while only 13 transcripts were not assembled Sensitivity of the indicated assembly methods at the locus level. Transcriptome assembled by the given method in the HAP1 dataset was compared against reference annotation (gencode hg38v35) using gffcompare.…”

Accurate isoform discovery with IsoQuant using long reads

“…Long-read RNA sequencing is now widely used in bulk, sorted cells, single cells and spatial approaches. This wide field of applications has led to the development of multiple spliced alignment programs [1][2][3][4] , transcript discovery methods [5][6][7][8][9][10][11] , tools for transcript classification 12 , annotation 13 and visualization 14,15 . Additionally, several reference-free tools for RNA long-read correction and assembly have been developed 16,17 .…”

“…For known transcripts, IsoQuant has the highest F1-score (the harmonic mean of precision and recall) compared to TALON 7 , FLAIR 8 , Bambu 11 and StringTie 5 , but these advances are not dramatic (Supplementary Tables 1-3). However, IsoQuant produces novel transcripts https://doi.org/10.1038/s41587-022-01565-y is likely to be highly useful across many genes, including but not limited to low-expressed long-noncoding RNAs and marker genes of cell types.…”