deSALT: fast and accurate long transcriptomic read alignment with de Bruijn graph-based index

Abstract

Long-read RNA sequencing (RNA-seq) is promising to transcriptomics studies, however, the alignment of long RNA-seq reads is still non-trivial due to high sequencing errors and complicated gene structures. Herein, we propose deSALT, a tailored two-pass alignment approach, which constructs graph-based alignment skeletons to infer exons and uses them to generate spliced reference sequences to produce refined alignments. deSALT addresses several difficult technical issues, such as small exons and sequencing errors, which breakthroughs the bottlenecks of long RNA-seq read alignment. Benchmarks demonstrate that deSALT has a greater ability to produce accurate and homogeneous full-length alignments. deSALT is available at: https://github.com/hitbc/deSALT. Keywords long read alignment, RNA-seq, de Bruijn graph-based index 3 Background RNA sequencing (RNA-seq) has become a fundamental approach to characterize transcriptomes.It reveals precise gene structures and quantifies gene/transcript expressions [1][2][3][4][5] in various applications, such as variant calling [6], RNA editing analysis [7, 8], and gene fusion detection [9, 10].However, current widely used short read sequencing technologies have limited read length and systematic bias from library preparation. These drawbacks limit more accurate alignment [11] and precise gene isoform analysis [12], thus creating a bottleneck for transcriptomic studies.Two kinds of long read sequencing technologies, i.e., single molecule real time (SMRT) sequencing produced by Pacific Biosciences (PacBio) [13] and nanopore sequencing produced by Oxford Nanopore Technologies (ONT) [14], are emerging and promising to breakthrough the bottleneck of short reads in transcriptomic analysis. Both of them enable the production of much longer reads, the mean and maximum lengths of the reads being over ten to hundreds of thousands of base pairs (bp) [15,16], respectively. Taking this advantage, full-length transcripts can be sequenced by single reads, which is promising for substantially improving the accuracy of gene isoform reconstruction. Furthermore, there is less systematic bias in the sequencing procedure [17], which is also beneficial to gene/transcript expression quantification.Besides their advantages, PacBio and ONT reads have much higher sequencing error rates than that of short reads. For PacBio SMRT sequencing, the sequencing error rate of raw reads ("subreads")is about 10% to 20% [16]; for ONT nanopore sequencing, the sequencing error rates of 1D and 2D (also known as 1D 2 ) reads are about 25% and 12% [18,19], respectively. PacBio SMRT platforms can produce reads of inserts (ROIs) by sequencing circular fragments multiple times to largely reduce sequencing errors. However, this technology has lower sequencing yields and reduced read lengths. Therefore, these high sequencing errors raise new technical challenges for RNA-seq data analysis.Read alignment could be the most affected one, and the effect may not be limited to the read alignment itself since it is fundamental to many down...

Nanopore sequencing technology, bioinformatics and applications

“…289 ), NanoBLASTer 290 , mapAlign 291 , GraphAligner 292 , smsmap 293 , lordFAST 294 , S-conLSH 295 , QAlign 296 Splice-aware minimap2, Graphmap2 (ref. 102 ), GmAP 100 , STAR 101 , deSALT 103 , magic-BLAST 297 , Deep-Long 298 , uLTRA 299 Genome assembly Canu, miniasm 107 , Flye 110 , Redbean/wtdbg2 (ref. 111 ), Falcon-Unzip 300 , Shasta 164 , Raven 301 , NextDenovo (https://github.com/Nextomics/NextDenovo), Peregrine 302 , HINGe 303 , TULIP 304 , NeCAT 305 metagenome tailored metaFlye 306 , OPeRA-mS (hybrid) 106 Haplotype-aware Hifiasm 307 Genome polishing Nanopolish, Racon 308 , medaka (https://github.com/nanoporetech/medaka), NeuralPolish 309 , PePPeR-margin-DeepVariant 310 , NextPolish (https://github.com/Nextomics/NextPolish), POLCA 311 , HomoPolish 38 SV detection Sniffles 98 , SVIm 312 , NanoSV 112 , Picky 33 , NanoVar 113 , Dysgu 313 , SeNSV 314 , cuteSV 315 SNV detection LongShot 116 , DeepVariant 310 , iGDA 316 , Nanopanel2 (ref.…”

“…Other aligners have also been developed, such as Graphmap2 (ref. 102 ) and deSALT 103 , for ONT transcriptome data. Especially for ONT direct RNA-sequencing reads with dense base modifications, Graphmap2 has a higher alignment rate than minimap2 (ref.…”

Sequencing of individual barcoded cDNAs on Pacific Biosciences and Oxford Nanopore reveals platform-specific error patterns

“…Sequence comparison using reference-based and reference-free alignments. Our further analysis mostly relies on read mappings, so at first we aligned reads using different tools: widely popular minimap2 (Li 2018) and specialized transcriptome aligners deSALT (Liu et al 2019) and GraphMap2 (Marić et al 2019) to exclude a possible bias. Although all three aligners yield largely similar results, GraphMap2 produced slightly shorter alignments than minimap2 and deSALT with more prominent differences between aligned lengths of PacBio and ONT reads.…”

“…Alignment to the genome using deSALT. PacBio reads were mapped to GRCm38 mouse reference genome with deSALT v1.5.6 (Liu et al 2019) using -t 16 -x ccs options. ONT reads were mapped with -t 16 -x ont1d options.…”

TALC: Transcript-level Aware Long Read Correction

“…In many cases these higher error rates can prevent the correct identification of isoforms (11)(12)(13). Although several alignment software (14)(15)(16)(17)(18) are optimized to handle these errors, their shortcomings confound transcript identification and annotation. Many reads cannot be aligned and regions where the sequencing error rates are higher such as UTRs frequently produce ambiguous alignments.…”