SNP Calling in the Zebrafish Mutation Project

The Zebrafish Mutation Project (ZMP) aims to isolate and functionally characterise a knock-out allele in every protein-coding gene in the zebrafish genome. The exomes of ENU-mutagenised F1 fish are screened by Illumina sequencing to identify induced nonsense and essential splice-site mutations. A high-throughput phenotyping pipeline is then used to link embryonic phenotypes to the identified genotypes.

SNP/insertion/deletion calling is done using a software pipeline developed at the Sanger Institute. To be considered, SNPs have to be called by all three SNP callers and pass a set of empirically-determined quality thresholds. This filtering produces SNP calls with a low false positive rate (~5%). To improve the filtering criteria we are analysing a subset of samples at relaxed stringency. This data will be used as a training set to implement a machine learning approach, already working well for Xenopus data, to improve our recovery of true positives while not significantly increasing the false positive rate.

We have so far identified 16149 nonsense and essential splice-site alleles in 9634 genes (37% of protein-coding genes) from 1554 F1 individuals. Based on the discovery rate of missense alleles, we predict that ~8000 individuals will be required to identify a nonsense or essential splice mutation in 75% of protein-coding genes. To compliment this we are producing Transcription Activator-Like Effector Nucleases (TALENs) as an additional strategy to target and disrupt specific genes. All the alleles identified are displayed on the ZMP website (http://www.sanger.ac.uk/Projects/D_rerio/zmp/) and distributed to external requestors.