The nature of TALEN and CRISPR/Cas9 lesions
Each of these systems similarly mutagenize their targets: by introducing double-stranded breaks at a targeted site. Imperfections in the endogenous repair process, known as non-homologous end-joining (NHEJ), leads to the formation of multiple alleles, typically in-dels ranging from 1 to 100 bp. It is worth noting that attempts to induce homologous recombination (HR) in zebrafish are ongoing in the research community, and several successes have been reported.
Consider the domain structure of the protein and alternate transcripts
Alleles with in-dels causing a reading frame shift predicted to scramble downstream code are generally expected to be null. For this reason researchers often target lesions upstream to domains with suspected essential biological activity. An exception to this principle is when nonsense-mediated decay can be succcesfuly induced, thereby disrupting both upstream and downstream sequences. Thus nonsense-mediated decay can add flexibility in target selection. Regardless of design, one must be vigilant and prepared for genetic robustness: the embryo's uncanny ability to "get by" with remaining genes (i.e., paralogs), alternate transcripts, alternate splice sites or alternate translational start sites.
To visualize the domain structure of the protein, use a tool like CD-Search. It is also advisable to select an exon that is required for all alternate transcripts. For finding alternate transcripts, the most up-to-date curations for zebrafish genes can be found within Ensembl's Vega site. If no Vega annotation is available, check Ensembl's main zebrafish browser, but be aware that a number of the transcripts reported here are "false positives" that will be removed through the manual annotation process.
Confirm your sequence
Zebrafish tend to have many SNPs, as they are not highly inbred. Indeed, the best breeding lines tend to have the most SNPs. One consequence of this is that the sequence you use to design a gene disruption project might have differences from the sequence in the embryos you select for disruption. If these differences lie within the recognition sites of TALENs or CRISPR guide RNAs, this can decrease mutational efficiency. The best insurance against this is to sequence your region of interest in the strain of embryos you intend to inject. Some researchers even take the precaution of sequencing each line or even each potential parent from which the embryos will be obtained. Although the above is advisable, a growing set of SNP databases can also help guide researchers towards trouble-free target regions. The NICHD Zebrafish core has access to some proprietary SNP databases, so please check with us for this step.