Development of a rare disease algorithm to identify persons at risk of Gaucher disease using electronic health records in the United States.

BACKGROUND: Early diagnosis of Gaucher disease (GD) allows for disease-specific treatment before significant symptoms arise, preventing/delaying onset of complications. Yet, many endure years-long diagnostic odysseys. We report the development of a machine learning algorithm to identify patients with GD from electronic health records. METHODS: We utilized Optum's de-identified Integrated Claims-Clinical dataset (2007-2019) for feature engineering and algorithm training/testing, based on clinical characteristics of GD. Two algorithms were selected: one based on age of feature occurrence (age-based), and one based on occurrence of features (prevalence-based). Performance was compared with an adaptation of the available clinical diagnostic algorithm for identifying patients with diagnosed GD. Undiagnosed patients highly-ranked by the algorithms were compared with diagnosed GD patients. RESULTS: Splenomegaly was the most important predictor for diagnosed GD with both algorithms, followed by geographical location (northeast USA), thrombocytopenia, osteonecrosis, bone density disorders, and bone pain. Overall, 1204 and 2862 patients, respectively, would need to be assessed with the age- and prevalence-based algorithms, compared with 20,743 with the clinical diagnostic algorithm, to identify 28 patients with diagnosed GD in the integrated dataset. Undiagnosed patients highly-ranked by the algorithms had similar clinical manifestations as diagnosed GD patients. CONCLUSIONS: The age-based algorithm identified younger patients, while the prevalence-based identified patients with advanced clinical manifestations. Their combined use better captures GD heterogeneity. The two algorithms were about 10-20-fold more efficient at identifying GD patients than the clinical diagnostic algorithm. Application of these algorithms could shorten diagnostic delay by identifying undiagnosed GD patients.

Identification of predictive factors of diabetic ketoacidosis in type 1 diabetes using a subgroup discovery algorithm.

AIM: To identify predictive factors for diabetic ketoacidosis (DKA) by retrospective analysis of registry data and the use of a subgroup discovery algorithm. MATERIALS AND METHODS: Data from adults and children with type 1 diabetes and more than two diabetes-related visits were analysed from the Diabetes Prospective Follow-up Registry. Q-Finder, a supervised non-parametric proprietary subgroup discovery algorithm, was used to identify subgroups with clinical characteristics associated with increased DKA risk. DKA was defined as pH less than 7.3 during a hospitalization event. RESULTS: Data for 108 223 adults and children, of whom 5609 (5.2%) had DKA, were studied. Q-Finder analysis identified 11 profiles associated with an increased risk of DKA: low body mass index standard deviation score; DKA at diagnosis; age 6-10 years; age 11-15 years; an HbA1c of 8.87% or higher (≥ 73 mmol/mol); no fast-acting insulin intake; age younger than 15 years and not using a continuous glucose monitoring system; physician diagnosis of nephrotic kidney disease; severe hypoglycaemia; hypoglycaemic coma; and autoimmune thyroiditis. Risk of DKA increased with the number of risk profiles matching patients' characteristics. CONCLUSIONS: Q-Finder confirmed common risk profiles identified by conventional statistical methods and allowed the generation of new profiles that may help predict patients with type 1 diabetes who are at a greater risk of experiencing DKA.

APPINetwork: an R package for building and computational analysis of protein-protein interaction networks.

Background: Protein-protein interactions (PPIs) are essential to almost every process in a cell. Analysis of PPI networks gives insights into the functional relationships among proteins and may reveal important hub proteins and sub-networks corresponding to functional modules. Several good tools have been developed for PPI network analysis but they have certain limitations. Most tools are suited for studying PPI in only a small number of model species, and do not allow second-order networks to be built, or offer relevant functions for their analysis. To overcome these limitations, we have developed APPINetwork (Analysis of Protein-protein Interaction Networks). The aim was to produce a generic and user-friendly package for building and analyzing a PPI network involving proteins of interest from any species as long they are stored in a database. Methods: APPINetwork is an open-source R package. It can be downloaded and installed on the collaborative development platform GitLab (https://forgemia.inra.fr/GNet/appinetwork). A graphical user interface facilitates its use. Graphical windows, buttons, and scroll bars allow the user to select or enter an organism name, choose data files and network parameters or methods dedicated to network analysis. All functions are implemented in R, except for the script identifying all proteins involved in the same biological process (developed in C) and the scripts formatting the BioGRID data file and generating the IDs correspondence file (implemented in Python 3). PPI information comes from private resources or different public databases (such as IntAct, BioGRID, and iRefIndex). The package can be deployed on Linux and macOS operating systems (OS). Deployment on Windows is possible but it requires the prior installation of Rtools and Python 3. Results: APPINetwork allows the user to build a PPI network from selected public databases and add their own PPI data. In this network, the proteins have unique identifiers resulting from the standardization of the different identifiers specific to each database. In addition to the construction of the first-order network, APPINetwork offers the possibility of building a second-order network centered on the proteins of interest (proteins known for their role in the biological process studied or subunits of a complex protein) and provides the number and type of experiments that have highlighted each PPI, as well as references to articles containing experimental evidence. Conclusion: More than a tool for PPI network building, APPINetwork enables the analysis of the resultant network, by searching either for the community of proteins involved in the same biological process or for the assembly intermediates of a protein complex. Results of these analyses are provided in easily exportable files. Examples files and a user manual describing each step of the process come with the package.

SURFMAP: A Software for Mapping in Two Dimensions Protein Surface Features.

Molecular cartography using two-dimensional (2D) representation of protein surfaces has been shown to be very promising for protein surface analysis. Here, we present SURFMAP, a free standalone and easy-to-use software that enables the fast and automated 2D projection of either predefined features of protein surface (i.e., electrostatic potential, hydrophobicity, stickiness, and surface relief) or any descriptor encoded in the temperature factor column of a PDB file. SURFMAP proposes three different "equal-area" projections that have the advantage of preserving the area measures. It provides the user with (i) 2D maps that enable the easy and visual analysis of protein surface features of interest and (ii) maps in a text file format allowing the fast and straightforward quantitative comparison of 2D maps of homologous proteins.

Can whole-exome sequencing data be used for linkage analysis?

Whole-exome sequencing (WES) has become the strategy of choice to identify causal variants in monogenic disorders. However, the list of candidate variants can be quite large, including false positives generated by sequencing errors. To reduce this list of candidate variants to the most relevant ones, a cost-effective strategy would be to focus on regions of linkage identified through linkage analysis conducted with common polymorphisms present in WES data. However, the non-uniform exon coverage of the genome and the lack of knowledge on the power of this strategy have largely precluded its use so far. To compare the performance of linkage analysis conducted with WES and SNP chip data in different situations, we performed simulations on two pedigree structures with, respectively, a dominant and a recessive trait segregating. We found that the performance of the two sets of markers at excluding regions of the genome were very similar, and there was no real gain at using SNP chip data compared with using the common SNPs extracted from WES data. When analyzing the real WES data available for these two pedigrees, we found that the linkage information derived from the WES common polymorphisms was able to reduce by half the list of candidate variants identified by a simple filtering approach. Conducting linkage analysis with WES data available on pedigrees and excluding among the candidate variants those that fall in excluded linkage regions is thus a powerful and cost-effective strategy to reduce the number of false-positive candidate variants.