The story continues. This week, I am trying to catch up with a number of recent papers in the field of neurogenetics. A recent publication in Nature Genetics highlights the role of de novo mutations identified through exome sequencing in schizophrenia. The authors also look at control data and compare their findings with the growing body of data available for autism research. And while many aspects regarding de novo mutations become more clear with every study published, the real difference is sometimes difficult to grasp.
Trio sequencing in schizophrenia. The recent study by Xu and colleagues looks a de novo mutations in a cohort of 231 trios with schizophrenia and 34 unaffected trios. The authors performed exome sequencing and looked a de novo mutations. The control trios provide an interesting baseline to compare the schizophrenia trios against. In the 34 unaffected trios, the authors observed 16 exonic de novo mutations, 11 of which were non-synonymous. Based on this data, a point mutation rate of 1.3 x 10 e-8 and nonsynonymous-to-synonymous ratio of 2.2 can be assessed. Particularly for the latter, control data is available, which suggests a similar rate. This translates into 0.5 exonic mutations per trio.
Measuring the difference. In the schizophrenia trios, the point mutation rate was not significantly different from controls, suggesting that the potential increase in de novo burden does not stand out against the genomic noise of random de novo mutations. However, the authors find a strong increase in the nonsynonymous-to-synonymous ratio in the schizophrenia trios. This suggests that the mutations found in patients with schizophrenia are more likely to affect function. This effect was even more extreme when the authors only considered variants that were very likely to be pathogenic, e.g. stop-mutations or frameshifts. However, given the size and effort of the study, you might be surprised by the unimpressive p-values for these comparisons, which are often in the 0.03 range and barely significant.
Numbers to remember. Based on their findings, the authors estimate the percentage of genes in patients with de novo mutation that are thought to be pathogenic. This percentage is in the range of 46%, i.e. every second mutation found can be expected to contribute to the phenotype. Likewise, the authors try to estimate the overall number of schizophrenia genes based on their data. Their estimate is ~800.
Enrichment. The functional annotation tool DAVID was used by the authors to find a common denominator between the de novo mutations. The authors did not find any convincing enrichment, demonstrating either the problem of noise or the difficulty to connect and discern yet unknown pathways. The authors identified gene clusters around MTOR and CANX as two potential candidates in a secondary protein-protein network analysis.
The prenatal story. The authors continued to focus on genes with a predominantly prenatal gene expression. They observed an even higher nonsynonymous-to-synonymous ratio. In addition, most genes overlapping with autism were from this group. The authors also commented that patients with de novo mutations in prenatally expressed genes have a more severe phenotype. While much of this data also is secondary analysis, it will be interesting to see whether a similar effect can also be found in further studies.
The recurrent genes and candidate genes. The authors find four genes (LAMA2, DPYP, TRRAP and VPS39) that are affected by de novo mutations in more than one patient, suggesting that these genes may represent genuine candidates given that a de novo mutation in two genes is unlikely to occur by chance. Also, the authors find two genes in the 22q region in patients with a known 22q11.2 microdeletion. To my knowledge, this is the first indication that some variants can be found “on the other strand” in patients with microdeletions.
Implications for EuroEPINOMICS. The growing data of sequencing studies in trios provides us with a good benchmark on where to place our epilepsy trios. This also includes a long list of >1000 genes known to be mutated in autism, schizophrenia or other neurodevelopmental disorders. It will be interesting to see whether we observe a similar overlap with other neurodevelopmental disorders as is the case for microdeletions.