Predict splicing variant effect from VCF
Paper: Cheng et al. https://doi.org/10.1101/438986
pip install mmsplice
Standard human gene annotation file in GTF format can be downloaded from ensembl or gencode.
MMSplice
can work directly with those files, however, some filtering is higly recommended.
- Filter for protein coding genes.
- Filter out duplicated exons. The same exon can be annotated multiple times if it appears in multiple transcripts. This will cause duplicated predictions.
We provide a filtered version here.
Note this version has chromosome names in the format chr*
. You may need to remove them to match the chromosome names in your fasta file.
A correctly formatted VCF file with work with MMSplice
, however the following steps will make it less prone to false positives:
- Quality filtering. Low quality variants leads to unreliable predictions.
- Avoid presenting multiple variants in one line by splitting them into multiple lines. Example code to do it:
bcftools norm -m-both -o out.vcf in.vcf.gz
- Left-normalization. For instance, GGCA-->GG is not left-normalized while GCA-->G is. Details for unified representation of genetic variants see Tan et al.
bcftools norm -f reference.fasta -o out.vcf in.vcf
Human reference fasta file can be downloaded from ensembl/gencode. Make sure the chromosome name matches with GTF annotation file you use.
Check notebooks/example.ipynb
To score variants (including indels), we suggest to use primarily the deltaLogitPSI
predictions, which is the default output. The differential splicing efficiency (dse) model was trained from MMSplice modules and exonic variants from MaPSy, thus only the predictions for exonic variants are calibrated.
# Import
from mmsplice.vcf_dataloader import SplicingVCFDataloader
from mmsplice import MMSplice, predict_all_table
from mmsplice.utils import max_varEff
# example files
gtf = 'tests/data/test.gtf'
vcf = 'tests/data/test.vcf.gz'
fasta = 'tests/data/hg19.nochr.chr17.fa'
gtfIntervalTree = 'tests/data/test.pkl' # pickle exon interval Tree
# dataloader to load variants from vcf
dl = SplicingVCFDataloader(gtf,
fasta,
vcf,
out_file=gtfIntervalTree, # same pikled gtf IntervalTree
split_seq=False)
# Specify model
model = MMSplice(
exon_cut_l=0,
exon_cut_r=0,
acceptor_intron_cut=6,
donor_intron_cut=6,
acceptor_intron_len=50,
acceptor_exon_len=3,
donor_exon_len=5,
donor_intron_len=13)
# Do prediction
predictions = predict_all_table(model, dl, batch_size=1024, split_seq=False, assembly=False)
# Summerize with maximum effect size
predictionsMax = max_varEff(predictions)
The VEP plugin wraps the prediction function from mmsplice
python package. Please check documentation of vep plugin under VEP_plugin/README.md.