Participants and cohort population stratification

One hundred twenty-one patients (P.1–P.121) met the study inclusion criteria listed in the Materials and Methods section. Their detailed phenotypes were analyzed and are presented in Table S1a. After accurate analysis and dynamic observation, some of the enrolled IGE diagnoses were questioned, and new possible diagnoses were added in brackets next to the previous ones. The diagnoses spectrum of the cohort (including questioned diagnoses) was as follows: 44 (36.4%) participants had JME, 30 (24.8%) had GTCSA, 27 (22.3%) had CAE, 18 (14.9%) had JAE, and two (1.7%) had unspecified IGE. Four (3.3%) of these diagnoses were questioned but were included in the study (may represent a common misdiagnosis rate in clinical practice).

The patient group comprised 75 children (62.0%) and 46 adults (38.0%). The mean age of the patients at the time of referral for WES was 16.7 ± 8.0 years (ranging from three to 40 years old). The patients consisted of 37 males (30.6%) and 84 females (69.4%). The detailed characteristics of our cohort are summarized in Table 1. All individuals in this cohort were from the Volga Federal District.

Table 1 Idiopathic generalized epilepsy (IGE) cohort characteristics

The observed population stratification in our cohort was estimated using PCA (Fig. 2), and the results were used to select the most reliable population frequencies in publicly available population databases (such as GnomAD) for further comparison. As shown in Fig. 2, the case cohort in the PCA figure was located between European and South Asian populations, partly overlapping with the Finnish population. Using merged data of the IGE sample with only the European populations, we found that our cohort had a unique genetic background within the European populations (Figure S1).

Fig. 2

Principal component analysis of IGE case samples with all 1KG populations. AFR African (ACB African-Caribbean, ASW African-American, ESN Esan, GWD Gambian, LWK Luhya, MSL Mende, YRI Yoruba), AMR American (CLM Colombian, MXL Mexican-American, PEL Peruvian, PUR Puerto Rican), EAS East Asian (CDX Dai Chinese, CHB Han Chinese, CHS Southern Han Chinese, JPT Japanese, KHV Kinh Vietnamese), EUR European (CEU CEPH, FIN Finnish, GBR British, IBS Spanish, TSI Tuscan), SAS South Asian (BEB Bengali, GIH Gujarati, ITU Indian, PJL Punjabi, STU Sri Lankan)

Candidate variants in epilepsy panel genes

After performing high-quality WES (mean coverage in the cohort was 125.6 ± 19.9, mean mapping quality – 56.6 ± 0.2, mapped reads – 99.5%), we obtained 39726 QVs, 1710 of them in the Genes4Epilepsy curated panel genes. The subsequent analyses revealed 168 CVs in 88 of 121 individuals (72.7%): 133 nonsynonymous, 16 loss-of-function SNVs and 19 frameshift indels (Table S2). For 33 out of 121 participants (27.3%) we could not identify any CVs. Of the remaining 88 participants, 51 (42.2%, 51/121) had two or more CVs. The distribution of all CVs per patient is shown in Table S5. All but three recurrent CVs in CPT2, DMXL2 and PINK1 genes were either not observed or had no frequency information available in the RuSeq control cohort.

Of the 168 CVs, 61 variants (36.3%) were novel variants (39 nonsynonymous SNVs, 8 loss-of-function SNVs and 14 frameshift indels), while 60 variants (35.7%) were reported in the ClinVar database (Table S2). Among the CVs reported in the ClinVar database, 10 LP nonsynonymous SNVs, 12 P loss-of-function SNVs and 15 P/LP frameshift indels were identified in 26.5% (32/121) of participants in 35 genes. These 37 CVs were found to be heterozygous in the affected individuals and were not observed in the control cohort.

All the CVs were heterozygous in the patients, except for one homozygous CV in DMXL2 (P.93, JME) (Table S2). Of note, among all CVs, five (4.1%, 5/121) variants were in five GGE-associated genes (CHD2, GABRA1, RORB, SCN1A, and SCN1B): one LP and three VUS nonsynonymous SNVs and one P loss-of-function SNV based on the ACMG guideline (Table 2). Mutations resulting in an amino acid change due to the combination of two nonsynonymous variants within a single codon were identified in HEXA (p.V363Y) and ANKRD17 (p.G445I), respectively (Table S2j).

Table 2 Candidate variants identified in GGE-associated genes in the idiopathic generalized epilepsy (IGE) cohortEpilepsy panel genes distribution in the cohort

Out of 950 candidate panel genes, 137 candidate genes (14.4%) had at least one variant that could contribute to the development of epilepsy in our case cohort (Table S3 and S4). For 76.6% of the candidate genes (105/137) with CVs, only one unique CV was identified in our cohort. The most frequently identified gene with CVs was SCN10A (4 variants in 4 individuals), followed by SACS (3 variants in 3 individuals). Five genes (CPT2, DMXL2, LYST, PIGW, and PINK1) had recurrent variants. No differences in the distribution pattern of candidate genes with CVs were observed among IGE syndromes or between familial and sporadic cases.

Family cases analysis

Twenty-five individuals had a known family history of epilepsy (including IGE, Rolandic epilepsy, and Panayiotopoulos syndrome). The WES data of their ten relatives (R.1–R.10) were separately analyzed for comparison with the probands’ CVs. Detailed phenotype descriptions of these 10 families are provided in Supplementary materials and Table S1b. We identified CVs shared by affected individuals in the candidate genes in each of four family cases (Table 3, Table S2).

Table 3 Shared candidate variants in probands and their relatives

Family P.15/R.1. Considering the autosomal dominant inheritance in the proband (3 years old boy, CAE) and his mother (29 years old, JME), we identified heterozygous CVs in HEXA (the combination of rs746202567 c.1088T>A and rs557550173 c.1087G>T, resulting in p.V363Y amino acid change) and TIAM1 (rs770266482 c.1481C>T; p.A494V).

Family P.22/R.4. Considering the autosomal recessive inheritance, we could not find any homozygous CVs in the proband (10-year-old girl, CAE) and her maternal uncle (33 years old, CAE). Considering the autosomal dominant inheritance with incomplete penetrance, we identified a heterozygous CV in SCN10A (rs145712124 c.2428G>T; p.G810W).

Family P.31/R.7. Considering the autosomal dominant inheritance in the proband (38-years old male, GTCSA) and his daughter (10 years old, Rolandic epilepsy), we identified a heterozygous CV in NDST1 (rs200193567 c.394C>T; p.R132C).

Family P.65/R.9: Considering the autosomal dominant inheritance in the proband (22 years old girl, JAE) and her father (45 years old, JME), we identified a previously unreported heterozygous CV in DENND5A (c.1739T>C; p.M580T).

Single cell type expression analysis of epilepsy panel genes

Based on the human single cell type transcriptomics map [33], the most frequent single cell type with the maximal expression level was “Excitatory neurons” cell type (“Neuronal cells” group) in the epilepsy panel genes: 13/137 (9.5%) genes in the “appearing genes” for which CVs were identified and 75/807 (9.3%) in the remaining “non-appearing genes” in our cohort (Table S6). Enrichment analysis for the single cell type with the maximal expression level revealed significantly higher fractions of “Astrocytes” (OR = 3.1, p = 1.0 × 10−2) and “Oligodendrocyte precursor cells” (OR = 2.9, p = 2.0 × 10−2) (“Glial cells” group) in the “appearing genes”, compared to the “non-appearing genes” (Fig. 3 and Table S6).

Fig. 3

Enrichment analysis of the epilepsy panel genes based on the single cell type transcriptomics map. Odds ratios (log10 scale), their confidential intervals (95%), and p-values to assess the differences in the proportions of the single cell types with the maximal expression levels between the “appearing genes” with CVs and the remaining “non-appearing genes” are shown for 38 single cell types with 2 or more “non-appearing genes”. The color of the circle represents a nominal p-value (-log10 transformed) from Fisher’s exact test. The size of the circle is proportional to the number of “appearing genes” genes that have the maximal expression levels for the single cell type