A subset of eDel loci were observed to harbor rare variants in both discovery and replication cohorts, but none of 2539 controls. eDel: exonic deletion; ACRD: autism chromosome rearrangement database (http://projects.tcag.ca/autism/).

APredicted event sizes in bps for unrelated AGRE* and ACC# cases are as follows: CA6 - 2317*, 12464*, 19146*, 19145#; OR1C1 - 44644*, 577691*, 1464677*, 44643#; NRXN1 - 19979*, 152437*, 241327*, 373015*, 439406*, 134010#, 161199#, 256373#, 533842#; GALNT13 - 14126*, 46413*, 113282*, 24100#; SUCLG2 - 2192*, 1389749*, 1389748#; KIAA1586 - 36354*, 36902*,157321*, 36901#, 67160#; RNF133/RNF148 - 33473*, 37226*, 1515817*, 43966#; SSSCA1/FAM89B – 21993*, 21993*, 21993*, 123569#; UNC93B1 - 11410*, 19223*, 84727*, 159861#; MDGA2 – 19651*, 23292*, 57714*, 58528*, 122985*, 131623*, 150178*, 226468*, 194601#, 288518#; LOC650137/OR4M2/OR4N4 – 591007* in 26 families, 24941# and 926360#; FLYWCH1/KREMEN2/PAQR4/PKMYT1 - 40468*, 81127*, 88373*, 82786#, 82786#, 124947#; BZRAP1 – 10102*, 10102*, 16897*, 18532*, 22806*, 34235*, 29600#, 28360#; C19orf19/MADCAM1 - 100187*, 171989*, 187147*, 98264#, 103788#, 277715#, 280201#, 292525#, 294446#, 344224#, 384324#.
BFor eDels at a given locus, the ratio of unaffected carriers (siblings or parents) to total number of carriers (cases and family members).
CThe significant difference in CNV frequency between AGRE and ACC cases (p = 2.6×10−6), along with multiple instances of similar variation in the DGV (see Tables S4 and Table S5), suggests that additional factors – including some potentially unrelated to diagnosis – may be relevant here. Sparse SNP coverage along with regional complexity (large segmental duplications) is also likely to increase false positive and false negatives at this locus. Replication data (and corresponding p Value) is for OR4N4, as only one eDel at either LOC650137 or OR4M2 was observed amongst ACC cases.
DA comparable number of eDels were observed at multiple neighboring genes; carrier fraction corresponds to FLYWCH1, the lowest observed at this locus.
EJoint consideration of eDels (n = 8) and eDups (n = 4) at BZRAP1 further improves statistical support for this locus (p = 2.3×10−5).
FNote extreme telomeric position of this locus which may undermine/interfere with reliable calling of structural variants. CNV counts and carrier fraction corresponds to MADCAM1; fewer variants were observed at C19orf19 amongst ACC cases and carrier fraction was higher than that for MADCAM1.