BAYAT eT Al | INTRODUCTIONGlycosylphosphatidylinositol (GPI) is a glycolipid that is synthetized and transferred to proteins in the membrane of the endoplasmic reticulum. 1 Biogenesis of GPI-anchored proteins (GPI-APs) is a conserved posttranslational mechanism in eukaryotes and is important for attaching these proteins to the cell membrane and for protein sorting, trafficking, and dynamics. 1,2 GPI synthesis and GPI-AP modification are mediated by at least 31 genes, and pathogenic variants in 22 of these genes have been associated with human disease to date. 3 The X-linked phosphatidylinositol glycan class A protein gene (phosphatidylinositol glycan class A protein [PIGA]) is part of a heptameric enzyme complex catalyzing the transfer of N-acetylglucosamine (GlcNAc) to phosphatidylinositol as the first step in GPI anchor biosynthesis. [4][5][6] In contrast to other members of the GPI-GlcNAc transferase complex, PIGA is an integral membrane protein with only one transmembrane domain residing in the endoplasmic reticulum (ER). The large N-terminal cytoplasmic domain contains two Rossmann folds. 7 Pathogenic germline missense variants in PIGA are associated with multiple congenital anomalies-hypotonia-seizures syndrome 2 (OMIM 316818). [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] The affected

The majority of monogenic disorders cause craniofacial abnormalities with characteristic facial morphology. These disorders can be diagnosed more e ciently with the support of computer-aided nextgeneration phenotyping tools, such as DeepGestalt. These tools have learned to associate facial phenotypes with the underlying syndrome through training on thousands of patient photographs. However, this "supervised" approach means that diagnoses are only possible if they were part of the training set. To improve recognition of ultra-rare diseases, we created GestaltMatcher, which uses a deep convolutional neural network based on the DeepGestalt framework. We used photographs of 21,836 patients with 1,362 rare disorders to de ne a "Clinical Face Phenotype Space". Distance between cases in the phenotype space de nes syndromic similarity, allowing test patients to be matched to a molecular diagnosis even when the disorder was not included in the training set. Similarities among patients with previously unknown disease genes can also be detected. Therefore, in concert with mutation data, GestaltMatcher could accelerate the clinical diagnosis of patients with ultra-rare disorders and facial dysmorphism.

HPMRS or Mabry syndrome is a heterogeneous glycosylphosphatidylinositol (GPI) anchor deficiency that is caused by an impairment of synthesis or maturation of the GPI‐anchor. The expressivity of the clinical features in HPMRS varies from severe syndromic forms with multiple organ malformations to mild nonsyndromic intellectual disability. In about half of the patients with the clinical diagnosis of HPMRS, pathogenic mutations can be identified in the coding region in one of the six genes, one among them is PGAP3. In this work, we describe a screening approach with sequence specific baits for transcripts of genes of the GPI pathway that allows the detection of functionally relevant mutations also including introns and the 5′ and 3′ UTR. By this means, we also identified pathogenic noncoding mutations, which increases the diagnostic yield for HPMRS on the basis of intellectual disability and elevated serum alkaline phosphatase. In eight affected individuals from different ethnicities, we found seven novel pathogenic mutations in PGAP3. Besides five missense mutations, we identified an intronic mutation, c.558‐10G>A, that causes an aberrant splice product and a mutation in the 3′UTR, c.*559C>T, that is associated with substantially lower mRNA levels. We show that our novel screening approach is a useful rapid detection tool for alterations in genes coding for key components of the GPI pathway.