**3. Copy number variations (CNVs)**

Dysregulation of gene expression of several genes/loci converging on the same networks (overlay) and/or combinatorial effects of different deleterious genetic variations appear to exceed a threshold and result in the autistic phenotype. In support of these ideas, strategies based on bioinformatics have identified many candidate genes, showing that ASD can be triggered by different types of genetic variations in many different genes, a phenomenon known as non-allelic genetic heterogeneity [74, 138, 105]. This model is more accepted; combines both common and rare variations posing risk for ASD, particularly those involving synaptic genes and genes involved in neurogenesis [138]. Thus it is assumed that people with ASD have a set of genetic variants that predispose them to abnormal development of brain structures involved in processing social information (the "social brain"). But it is known that there is no common pathophysiology in ASD. This may result from mutations in many different genes involved in different functions [138]. The kinds of variants that incline to autism and can involve several genes at the same time are the CNVs.

CNVs are microduplications or microdeletions resulting from insertions, deletions or trans‐ locations in the human genome that are observed in the general population and commonly found in genic regions in individuals with neuropsychiatric disorders. They can be inherited or *de novo*, frequent or rare with a frequency of less than 1% of the population. A substantial portion of autism cases appears to result from rare CNVs with variations larger than 100kb; they are more common in individuals with ASD than in the general population [33, 83, 118].

*De novo* CNVs have been reported in 5-10% of cases of idiopathic ASD. Many studies have revealed that some CNVs occur at significantly higher frequencies than others and some are exclusively observed subjects and not found in normal controls. This has allowed the identi‐ fication of new candidate genes which have not yet been described in the Autism Chromosome Rearrangement Database, such as *GABRA5, GABRA3, GABRG3, UBE3A, E2F1, PLCB1, PMP22,* *AADAT, MAPK3, NRXN1, NRG3, DPP10, UQCRC2, USH2A, NECAB3, CNTN4, LINGO2, IL1RAPL1, STXBP5, DOC2A, SNRPN, E2F1, AADAT, NECAB3, GPHN, dlg2, HPCAL1, BDNF-OS* and *IL1RAPL1*, (http://projects.tcag.ca/autism/). The new risk loci for ASD have functions that suggest an important role in the function and architecture of the brain; one CNV could interfere with normal biochemical pathways and predispose to the disorder [27, 46, 95). CNVs associated with ASD and schizophrenia are also associated with cognitive problems in control subjects showing a probable variable expressivity of these changes [134].

Although structural variations, such as CNVs, play an important etiologic role in the devel‐ opment of ASD as has been proposed by several authors since 2006, most of the results of different studies are not considered in the clinical evaluation of children with ASD, probably due to the rarity of individual variants, the lack of coverage of probes in clinical microarrays, the lack of reproducibility of studies that present different findings and the difficulty to understand the biology corresponding to some variants even when they are significantly associated with ASD. Nevertheless, clinical guidelines suggest that microarray-based tests are the first step in the genetic analysis of children with ASD [128, 68]; however this is not feasible in most cases of low-income countries due to the high cost of these tests.
