Skip to main content
. 2022 May 6;2(1):vbac032. doi: 10.1093/bioadv/vbac032

Table 1.

Comparison of splice variant prediction tools from the literature with NeoSplice

Splice variant antigen caller Input Predicts neoantigens Splicing event identified Required packages In silico performance Wet lab validation Wet lab performance
ASNEO RNA-seq Yes Filters reads against GTEx and hg19 reference, translating novel isoforms into proteins for antigen prediction.

  • Python: sj2psi

  • R: survival, survminer, MCPcounter

  • Not reported

 Mass spectrometry (external dataset) 2/407 peptides confirmed from 14 patient cohort
JuncBase RNA-seq No (1) Identifies annotated and novel splice junctions, (2) quantifies each junction and (3) calculated for differential expression between groups.

  • Python 2.6+

  • Biopython 1.5+

  • Pysam

  • R v2.14+

  • Rpy2

  • MySQL/sqlite

  • Sens. ∼50–80%

  • Prec. ∼10–95%

  • (Compared in Kahles et al.)

 RT-PCR 16/16 splicing events confirmed
MiSplice RNA-seq + WGS No Jointly analyzes WGS and RNA-Seq data, scanning the transcriptome for statistically significant non-canonical sequence junctions supported by expression evidence.

  • SamTools

  • MaxEntScan

  • Sens. 74–97%

  • Spec. ∼77%

 Splicing reporter minigene functional assay 10/11 of splicing alterations
MutPred Splice DNAseq No Uses human disease alleles for training a machine learning model to predict exonic nucleotide substitutions that disrupt pre-mRNA splicing.

  • None (web interface)

  • FPR = 7.0%

  • Sens. 64.7%

  • Spec. 93.0%

  • Acc. 78.8%

  • AUC 83.5%

 RT-PCR Amplicon changes from ATM mutation-contain vs WT cell line confirmed by RT-PCR
NeoSplice RNA-seq Yes (1) Identify differentially expressed k-mers, (2) map tumor-specific k-mers to splice graph and (3) ORF inference, translation, and MHC binding prediction.

  • Python 2.7

  • MSBWT

  • MSBWT-IS

  • NetMHCpan 4.0

  • NetMHCIIpan 3.2

  • networkx 1.11

  • pyahocorasick 1.4.0

  • bcbio-gff 0.6.4

  • pyfaidx 0.5.3.1

  • pysam 0.14.1

  • biopython 1.70

  • scipy 1.2.0

  • Sens. >80%

  • Recall >80%

 Internal mass spectrometry validation against synthetic peptide reference 4/37 peptides confirmed, corresponding to 3/17 novel splice junctions
RI neoantigen pipeline RNA-seq Yes (1) Pseudoaligns RNAseq reads to hg19 with exon and intron transcripts, (2) quantification, (3) KMA algorithm to identify expresed introns and (4) predict MHC binding.

  • Kallisto

  • KMA suite of python and R packages

  • POLYSOLVER

  • NetMHCpan v3.1

  • Not reported

 Mass spectrometry (external dataset) Confirmed 1–2 per each of six cell line tested (Mean total splice variant neantigen load of 1515)
rMATS RNA-seq No Detection of differentially expressed splice variants between two sets of RNA-seq data.

  • Python 2.7/3.6

  • BLAS, LAPACK

  • GSL 2.5

  • GCC (5.4.0)

  • Fortran 77

  • CMake (3.15.4)

  • Sens. ∼30%

  • Prec. >95%

  • (Compared in Kahles et al. for exon skipping only)

 RT-PCR 32/34 exon skipping candidates confirmed
SplAdder RNA-seq No (1) Integrating annotation and RNA-Seq data, (2) generating an augmented splicing graph, (3) extraction of splicing events, (4) quantifying the events, and optionally and (5) the differential analysis between samples.

  • LIMIX

  • GATK

  • STAR

  • Samtools

  • Sens. ∼30–80%

  • Prec. ∼20–90%

 None NA
SpliceGrapher

  • RNA-seq

  • +/-EST

 No (1) Alignment of RNA-seq to the reference genome, (2) spliced alignment of reads that did not align in the first step, (3) initial splice graph construction, (4) assembly of exons from the ungapped short-read alignments and (5) insertion of the new exons into the splice graph using spliced alignments.

  • PyML 0.7.9+

  • matplotlib 1.1.0+

  • pysam 0.5+

  • Sens. ∼30–60%

  • Prec. ∼20–90%

  • (Compared in Kahles et al.)

 None NA