Résumé : Early-onset myopathies are genetically heterogeneous mendelian diseases. We have performed solo or trio whole exome sequencing (WES) in a cohort of 24 individuals with genetically undiagnosed early-onset myopathies from 21 families. We have identified single nucleotide variant (SNV) or copy number variants (CNV) in known genes in 52% of the patients. We contributed to the expansion of the phenotypic spectrum for two genes (TTN, TRIP4). In particular, through international collaborative efforts, we described the severe end of the phenotypic spectrum of congenital titinopathy. Fifteen cases, with biallelic TTN truncating or canonical splice site mutations, presented with severe in utero features and/or severe weakness and respiratory compromise at birth, leading to termination of pregnancy, neonatal death or survival with long term respiratory support. The presence of multiple congenital limb contractures, congenital long bone fracture and/or complete absence of muscle groups can be diagnostic clues. All severely affected patients that survived until birth had at least one mutation predicted to result in the production of at least some near-normal-length titin, a finding that seems to be essential for human in utero survival.WES also allowed us to identify three novel candidate genes (JAG2, R....., U.....). Through Matchmaker Exchange, we assembled a cohort of 23 individuals with undiagnosed muscular dystrophy from 13 unrelated families with rare homozygous or compound heterozygous variants in JAG2. Individuals with JAG2 variants had a wide spectrum of phenotypes and disease severity but the muscle MRI findings appear to be the unifying feature that helps to identify these patients. Transcriptome analysis in the muscle of two affected individuals suggests satellite cell depletion, a finding that is also present in other muscular dystrophies linked to the Notch pathway like POGLUT1 or MEGF10 related muscular dystrophies. If the implication of our three novel candidate genes is confirmed, the diagnostic yield in the cohort would increase to 67%. The zebrafish has been widely used to validate the implication of novel genes of myopathy, mostly with morpholino or mutated mRNA injections. As the specificity of morpholinos and mutated mRNA injections had been questioned, we have decided to use CRISPR-Cas9 to model loss-of-function or gain-of-function by creating knockout or knockin lines. We generated a knockout line for mtm1 and a knockin line for the S619L gain-of-function mutation in dnm2a because these mutations were associated with severe myopathic phenotypes in humans and had already been modeled in zebrafish with morpholino or mutated mRNA injections. In contrast to the severe phenotype of morpholino/mRNA injected fish, we did not observe any structural or functional defect in the muscle of mtm1 mutants and dnm2a mutants. Off-target effects of morpholino or mRNA injections as well as genetic compensation mechanisms in mutant lines are possible mechanisms explaining these discrepancies. The absence of phenotype in a zebrafish mutant line is not sufficient to rule out the implication of a gene in a human phenotype.