Table 1 Key supporting evidence for the contribution of de novo mutations to the genetic aetiology of ASD.

Sebat et al. (2007) [23]

264 families (118 simplex, 47 multiplex, 99 control)

Comparative genomic hybridization

The frequency of de novo CNVs was significantly higher in ASD-affected relative to unaffected children, and nominally higher in simplex relative to multiplex families

Marshall et al. (2008) [22]

427 ASD families (237 simplex, 189 multiplex)

Microarray analysis & karyotyping

De novo CNVs were more frequently detected among ASD cases from simplex families compared to those from multiplex families, however both exceeded the de novo rate expected in control families. Karyotyping revealed a small number of additional balanced cytogenic anomalies of de novo origin among ASD probands (simplex and multiplex)

Pinto et al. (2010) [118]

996 ASD probands, 1287 matched controls

Genome-wide microarray analysis

There was a significant 1.19-fold increase in the burden of rare genic CNVs, including de novo variants, for ASD cases relative to controls. Over 5% of ASD cases possessed at least one de novo CNV, with >0.6% carrying two or more de novo variants

Levy et al. (2011) [119]

915 families (889 ASD probands, 895 control siblings)

Comparative genomic hybridization

De novo CNVs were detected in 7.9% of ASD probands versus 2.0% of all siblings, with CNVs occurring significantly more frequently in genic regions in ASD-affected compared to unaffected siblings

O’Roak et al. (2011) [120]

20 ASD trios

Exome sequencing

The observed rate of de novo protein coding mutations was slightly higher among ASD probands than expected in the general population. Potentially causative de novo mutations were detected in four of the most severely affected cases according to IQ and calibrated severity scores

Sanders et al. (2011) [121]

1124 families (872 quads, 8252 trios^b)

Genome-wide microarray analysis

De novo CNVs were significantly more frequent in ASD probands compared to unaffected siblings (OR = 3.5), with CNVs being larger on average and harbouring a larger number of genes

Iossifov et al. (2012) [122]

343 ASD quads^b

Exome sequencing

Likely gene-disrupting de novo mutations, including nonsense, splice site, and frameshift mutations were twice as frequent among ASD probands relative to control siblings

Sanders et al. (2012) [19]

225 families (200 quads, 25 trios^b)

Exome sequencing

The frequency of nonsynonymous de novo SNVs, including nonsense SNVs and splice site altering SNVs, were higher in affected relative to unaffected siblings, with a significant increase in the proportion of gene-disrupting compared to silent mutations among probands versus siblings (OR = 1.93)

Jiang et al. (2013) [123]

32 trios^b

Genome sequencing

De novo SNVs likely to contribute to ASD presentation were detected in 19% of examined probands, affecting both known and novel ASD risk genes

De Rubeis et al. (2014) [55]

2270 ASD trios^b, 510 control trios

Exome sequencing

De novo loss-of-function mutations were observed significantly more frequently in ASD trios relative to matched control trios

Dong et al. (2014) [124]

787 families (602 quads, 185 trios^b)

Exome sequencing

Spontaneous de novo frameshift indels were found to contribute to ASD risk among probands (OR = 1.6)

Iossifov et al. (2014) [12]

2517 families (2508 ASD probands, 1911 control siblings)

Exome sequencing

The ascertainment differentials for the rates of likely gene-disrupting and missense de novo mutations were significant when comparing ASD probands to unaffected siblings, contributing to 12% and 9% of ASD diagnoses, respectively

O’Roak et al. (2014) [125]

3486 ASD probands, 2493 control siblings

Exome sequencing

Observed an elevated rate of de novo mutations in candidate ASD genes among affected relative to unaffected siblings, with a 2.7-fold greater rate of protein-altering de novo mutations among ASD probands

Pinto et al. (2014) [21]

2446 ASD families (simplex & multiplex), 4768 control subjects

Genome-wide microarray analysis

Reported a significant 1.41-fold excess of genic de novo CNVs of greater average length among ASD cases compared to controls. ASD probands were more likely to possess de novo variation in ASD candidate genes and gene networks related to processes of neurodevelopment and gene regulation

Krumm et al. (2015) [126]

2377 families (2377 ASD probands, 1786 control siblings)

Exome sequencing

2.4-fold excess of de novo CNVs (deletions) were observed in ASD-affected relative to unaffected siblings, with CNVs in probands being significantly larger on average and enriched in variation-intolerant genes

Sanders et al. (2015) [127]

2591 families (2100 quads, 491 trios)

Genome-wide microarray analysis

The frequency, average size, and genic content of de novo CNVs was significantly higher among ASD cases relative to unaffected siblings

Leppa et al. (2016) [20]

1532 families (343 simplex, 1189 multiplex)

Genome-wide microarray analysis

Rare CNVs, including de novo events, were significantly more frequent among ASD-affected compared to unaffected siblings, as well as in simplex versus multiplex families, suggesting a higher burden of de novo variation in the genetic aetiology of sporadic ASD

Yuen et al. (2016) [40]

200 ASD trios^b

Genome sequencing

Relative to controls, ASD probands demonstrated enrichment of deleterious de novo mutations in non-coding expression control regions (5ʹ/3ʹUTR), splice sites predictive of exon skipping, and DNase I hypersensitivity regions

Turner et al. (2017) [128]

516 ASD quads^b

Exome sequencing

Significant enrichment for deleterious missense de novo SNVs/indels, promoter, and enhancer mutations were detected among ASD-affected relative to unaffected siblings

An et al. (2018) [129]

1902 ASD quads^b

Genome sequencing

De novo risk scores were found to significantly predict ASD case status when localised to promoter regions characterised by evolutionary and functional signatures