The UMB Digital Archive is getting an upgrade! The upgrade requires a content freeze starting 1/27/25 and is expected to last two weeks. Any new user accounts or submissions made to the Archive during this time will not be transferred to the upgraded site. Contact ArchiveHelp@hshsl.umaryland.edu for questions.
Evaluation of Kcnma1 Channelopathy Variants Using Combinatorial Pathogenicity Algorithms
Abstract
Mutations in the KCNMA1 gene that encodes the voltage and calcium gated potassium channel (BK) are linked to a neurological disorder characterized by seizures, movement disorder, and neurodevelopmental disability. Despite identifying some predicted and confirmed pathogenic variants, 41% remain variants of uncertain significance (VUS). To provide more detailed characterization for novel pathogenic variants, we tested the performance of the most widely used pathogenicity algorithms: Mutpred, CADD/PHRED, MetaLR, M-CAP and REVEL. Controls for channel-level pathogenicity in this analysis were the confirmed GOF mutations D434G and N999S and the LOF mutations G354S and D965V. Non-pathogenic variants K518N, E884K and R1128W comprised negative controls. Our results show that REVEL has the best performance to detect pathogenic and non-pathogenic mutations, with a true positive rate (TPR) of 0.86, true negative rate (TNR) of 0.80, and a logistic regression with the higher are under the curve (AUC=0.90). Additionally, correlation between REVEL and the other five algorithms improved their prediction performance, increasing the TPR and reducing the TNR based on the cut-off combination. Next, we combined these algorithms and generated a KCNMA1 Meta Score (KMS) incorporating structural information from the BK channel cryo-EM structures with and without Ca2+. The KMS algorithm performance curve shows 0.84 AUC with a cut-off associated with high TPR and low TNR, identifying several variants as pathogenic that were not predicted by REVEL, all together, these results demonstrate that the five algorithms show between them up to 7% differences in average performance that is translated in the ability to detect true pathogenic and non-pathogenic KCNMA1 variants. However, the incorporation of structural data results in a KMS score with similar to the best-ranked (REVEL), but with a small increase in the ability to detect pathogenic variants.Mutations in the KCNMA1 gene that encodes the voltage and calcium gated potassium channel (BK) are linked to a neurological disorder characterized by seizures, movement disorder, and neurodevelopmental disability. Despite identifying some predicted and confirmed pathogenic variants, 41% remain variants of uncertain significance (VUS). To provide more detailed characterization for novel pathogenic variants, we tested the performance of the most widely used pathogenicity algorithms: Mutpred, CADD/PHRED, MetaLR, M-CAP and REVEL. Controls for channel-level pathogenicity in this analysis were the confirmed GOF mutations D434G and N999S and the LOF mutations G354S and D965V. Non-pathogenic variants K518N, E884K and R1128W comprised negative controls. Our results show that REVEL has the best performance to detect pathogenic and non-pathogenic mutations, with a true positive rate (TPR) of 0.86, true negative rate (TNR) of 0.80, and a logistic regression with the higher are under the curve (AUC=0.90). Additionally, correlation between REVEL and the other five algorithms improved their prediction performance, increasing the TPR and reducing the TNR based on the cut-off combination. Next, we combined these algorithms and generated a KCNMA1 Meta Score (KMS) incorporating structural information from the BK channel cryo-EM structures with and without Ca2+. The KMS algorithm performance curve shows 0.84 AUC with a cut-off associated with high TPR and low TNR, identifying several variants as pathogenic that were not predicted by REVEL, all together, these results demonstrate that the five algorithms show between them up to 7% differences in average performance that is translated in the ability to detect true pathogenic and non-pathogenic KCNMA1 variants. However, the incorporation of structural data results in a KMS score with similar to the best-ranked (REVEL), but with a small increase in the ability to detect pathogenic variants.Description
Poster presented at SGP, September 7-11, 2022Rights/Terms
Attribution-NonCommercial-NoDerivatives 4.0 InternationalIdentifier to cite or link to this item
http://hdl.handle.net/10713/19652Collections
The following license files are associated with this item:
- Creative Commons
Except where otherwise noted, this item's license is described as Attribution-NonCommercial-NoDerivatives 4.0 International