Splice AI

Overview

SpliceAI, a 32-layer deep neural network, predicts splicing from a pre-mRNA sequence.

Publication

K. Jaganathan, et al. Predicting splicing from primary sequence with deep learning. Cell, 176 (3) (2019), pp. 535-548 e24

VCF File

Example

##fileformat=VCFv4.0
##assembly=GRCh37/hg19
##INFO=<ID=SYMBOL,Number=1,Type=String,Description="HGNC gene symbol">
##INFO=<ID=STRAND,Number=1,Type=String,Description="+ or - depending on whether the gene lies in the positive or negative strand">
##INFO=<ID=TYPE,Number=1,Type=String,Description="E or I depending on whether the variant position is exonic or intronic (GENCODE V24lift37 canonical annotation)">
##INFO=<ID=DIST,Number=1,Type=Integer,Description="Distance between the variant position and the closest splice site (GENCODE V24lift37 canonical annotation)">
##INFO=<ID=DS_AG,Number=1,Type=Float,Description="Delta score (acceptor gain)">
##INFO=<ID=DS_AL,Number=1,Type=Float,Description="Delta score (acceptor loss)">
##INFO=<ID=DS_DG,Number=1,Type=Float,Description="Delta score (donor gain)">
##INFO=<ID=DS_DL,Number=1,Type=Float,Description="Delta score (donor loss)">
##INFO=<ID=DP_AG,Number=1,Type=Integer,Description="Delta position (acceptor gain) relative to the variant position">
##INFO=<ID=DP_AL,Number=1,Type=Integer,Description="Delta position (acceptor loss) relative to the variant position">
##INFO=<ID=DP_DG,Number=1,Type=Integer,Description="Delta position (donor gain) relative to the variant position">
##INFO=<ID=DP_DL,Number=1,Type=Integer,Description="Delta position (donor loss) relative to the variant position">
#CHROM  POS ID  REF ALT QUAL    FILTER  INFO
10  92946   .   C   T   .   .   SYMBOL=TUBB8;STRAND=-;TYPE=E;DIST=-53;DS_AG=0.0000;DS_AL=0.0000;DS_DG=0.0000;DS_DL=0.0000;DP_AG=-26;DP_AL=-10;DP_DG=3;DP_DL=35
10  92946   .   C   G   .   .   SYMBOL=TUBB8;STRAND=-;TYPE=E;DIST=-53;DS_AG=0.0008;DS_AL=0.0000;DS_DG=0.0003;DS_DL=0.0000;DP_AG=34;DP_AL=-27;DP_DG=35;DP_DL=1
10  92946   .   C   A   .   .   SYMBOL=TUBB8;STRAND=-;TYPE=E;DIST=-53;DS_AG=0.0004;DS_AL=0.0000;DS_DG=0.0001;DS_DL=0.0000;DP_AG=-10;DP_AL=-48;DP_DG=35;DP_DL=-21
10  92947   .   A   C   .   .   SYMBOL=TUBB8;STRAND=-;TYPE=E;DIST=-54;DS_AG=0.0002;DS_AL=0.0000;DS_DG=0.0000;DS_DL=0.0000;DP_AG=-49;DP_AL=-11;DP_DG=0;DP_DL=34
10  92947   .   A   T   .   .   SYMBOL=TUBB8;STRAND=-;TYPE=E;DIST=-54;DS_AG=0.0002;DS_AL=0.0000;DS_DG=0.0000;DS_DL=0.0000;DP_AG=33;DP_AL=-11;DP_DG=-22;DP_DL=34
10  92947   .   A   G   .   .   SYMBOL=TUBB8;STRAND=-;TYPE=E;DIST=-54;DS_AG=0.0006;DS_AL=0.0000;DS_DG=0.0001;DS_DL=0.0000;DP_AG=33;DP_AL=-11;DP_DG=34;DP_DL=32

Parsing

From the VCF file, we're mainly interested in the following columns:

• DS_AG - Δ score (acceptor gain)
• DS_AL - Δ score (acceptor loss)
• DS_DG - Δ score (donor gain)
• DS_DL - Δ score (donor loss)
• DP_AG - Δ position (acceptor gain) relative to the variant position
• DP_AL - Δ position (acceptor loss) relative to the variant position
• DP_DG - Δ position (donor gain) relative to the variant position
• DP_DL - Δ position (donor loss) relative to the variant position

The Splice AI team suggests the following interpretation for the scores:

Range	Confidence	Pathogenicity
0 ≤ x < 0.1	low	likely benign
0.1 ≤ x ≤ 0.5	medium	likely pathogenic
x > 0.5	high	pathogenic

Pre-processing

Filtering

Splice AI provides a comprehensive list of entries throughout the genome. However, many of the entries have little value. I.e. observing low splice scores in intergenic regions. Not only do these extra entries require more storage, but the unused content has a negative impact on annotation speed.

As a result, Nirvana filters out all the values in the low confidence tier except for regions within 15 bp of nascent splice sites. For those regions, we found it useful to see if Splice AI predicted an interruption of the splicing mechanism.

Download URL

https://basespace.illumina.com/s/5u6ThOblecrh

JSON Output

"spliceAI":[ 
   {
      "hgnc":"BLCAP",
      "acceptorGainDistance":-3,
      "acceptorGainScore":0.3,
      "donorLossDistance":7,
      "donorLossScore":0.9
   },
   { 
      "hgnc":"NNAT",
      "acceptorGainDistance":-1,
      "acceptorGainScore":0.2,
      "donorGainDistance":-2,
      "donorGainScore":0.3
   }
]

Field	Type	Notes
hgnc	string	HGNC gene symbol
acceptorGainDistance	int	± bp from current position
acceptorGainScore	float	range: 0 - 1.0. 1 decimal place
acceptorLossDistance	int	± bp from current position
acceptorLossScore	float	range: 0 - 1.0. 1 decimal place
donorGainDistance	int	± bp from current position
donorGainScore	float	range: 0 - 1.0. 1 decimal place
donorLossDistance	int	± bp from current position
donorLossScore	float	range: 0 - 1.0. 1 decimal place