**Background:** Mendelian diseases affect all body systems and developmental stages, and despite advances in next-generation sequencing, diagnostic rates remain below 50% for exome sequencing and around 35% for whole-genome sequencing. Challenges in causal variant identification extend beyond technical coverage issues, yet large-scale systematic analyses of these pitfalls are lacking. This study aims to comprehensively characterize the challenges encountered in a large Mendelian genomics program involving 4577 molecularly characterized families.
**Methods:** The cohort comprised 4577 families (out of 8024 total) in which a likely causal variant was identified, with 94.5% from a research lab and ~96% from Saudi Arabia. Consanguinity (third cousin or closer) was documented in 81% of families. The cohort covered a broad range of disease pathologies including neurodevelopmental, dysmorphic/congenital malformation syndromes, inborn errors of metabolism, hematological, immunological, ophthalmological, audiological, pulmonary, gastrointestinal, connective tissue-related, cardiovascular, skeletal, reproductive, and renal. Age ranged from zygote to 80 years, with 51.8% males and 47.1% females. All families underwent genotyping and positional mapping; testing strategies included next-generation multi-gene panels, exome sequencing, chromosomal microarray, optical genome mapping, and other clinical tests. Variants were classified according to ACMG guidelines, and novel gene-disease assertions were classified using ClinGen guidelines. A detailed review of every family was undertaken to identify challenging scenarios, categorized into phenotype-related, gene-related, variant-related, pedigree-related, and positional mapping-related challenges.
**Key Results:** Among the 4577 families, 1570 (34.3%) had one or more challenges. The challenges were categorized as follows:
- **Phenotype-related:** Phenotypic heterogeneity (~3% of families), phenotypic expansion (79 families, 5%), novel allelic disorders (83 families, 5.3% representing 52 distinct allelic disorders), blended phenotypes (87 families, 5.5%), erroneous clinical labels (15 families, ~1%), and non-Mendelian phenotypes (28 families, 1.8%).
- **Gene-related:** Novel gene-disease assertions (132 not yet listed in OMIM, plus 155 with questionable assertion at time of analysis, total 9.8% of families), incompatible animal model phenotype (7 families, 0.45%), and known gene with novel mutation mechanism (71 families, 4.5%).
- **Variant-related:** Interpretation challenges included tentative transcript-deleterious variants (177 families, 11.3%), allele frequency above cut-off (255 families, 16.2%), challenging in silico prediction (70 families, 4.5%), complex compound inheritance (8 variants highlighted), multivariant alleles, incomplete penetrance (65 families, 4.1%), and distraction by other variants (240 families, 15.3%). Technical challenges included deep intronic variants, regulatory elements (e.g., a deletion affecting DKK1 enhancer in family F3029), repeat expansions, genomic rearrangements (42 families, 2.7%), pseudogenes (13 families, 0.8%), platform/bioinformatic limitations (68 families, 4.3%), and epigenetic changes.
- **Pedigree-related:** Pseudodominance (15 families, ~1%), gonadal mosaicism (7 families, 0.45%), parental balanced rearrangement (2 families, 0.13%), and intrafamilial genetic/allelic heterogeneity (140 families, 8.9%).
- **Positional mapping-related:** Lack of detectable ROH at disease locus (29 families, 1.9%), apparent sharing of ROH with unaffected (5 families, 0.32%), and double recombination (2 families).
- **Sample mix-up:** 6 families (0.38%).
Reanalysis of 314 families referred after negative clinical exome/genome sequencing identified a likely causal variant in 54.5%. The most common challenge in this subset was novelty of gene-disease assertion (48%), followed by variant-related challenges (37.4%), phenotype-related issues (11.7%), and pedigree-related challenges (1.8%). Only 15.2% of variants identified on reanalysis could not have been captured technically by exome sequencing.
**Clinical Implications:** This comprehensive analysis demonstrates that the majority of missed diagnoses in Mendelian disease genomics are due to interpretation challenges rather than technical limitations of sequencing. The study provides a framework for addressing these challenges, including the identification of 357 novel gene-disease assertions, 23 recessive forms of previously dominant-only genes, and 85 important founder variants in the local population. These findings have direct implications for improving diagnostic yield, refining variant interpretation, and enabling more accurate genetic counseling. The results argue for balanced investment in both sequencing technologies and interpretation pipelines to realize the full potential of clinical genomics.