AAgMarker 1.0: a resource of serological autoantigen biomarkers for clinical diagnosis and prognosis of various human diseases

Abstract

Autoantibodies are produced to target an individual's own antigens (e.g. proteins). They can trigger autoimmune responses and inflammation, and thus, cause many types of diseases. Many high-throughput autoantibody profiling projects have been reported for unbiased identification of serological autoantigen-based biomarkers. However, a lack of centralized data portal for these published assays has been a major obstacle to further data mining and cross-evaluate the quality of these datasets generated from different diseases. Here, we introduce a user-friendly database, AAgMarker 1.0, which collects many published raw datasets obtained from serum profiling assays on the proteome microarrays, and provides a toolbox for mining these data. The current version of AAgMarker 1.0 contains 854 serum samples, involving 136 092 proteins. A total of 7803 (4470 non-redundant) candidate autoantigen biomarkers were identified and collected for 12 diseases, such as Alzheimer's disease, Bechet's disease and Parkinson's disease. Seven statistical parameters are introduced to quantitatively assess these biomarkers. Users can retrieve, analyse and compare the datasets through basic search, advanced search and browse. These biomarkers are also downloadable by disease terms. The AAgMarker 1.0 is now freely accessible at http://bioinfo.wilmer.jhu.edu/AAgMarker/. We believe this database will be a valuable resource for the community of both biomedical and clinical research.

Human Autoantigen Atlas: Searching for the Hallmarks of Autoantigens

“…These data were integrated into our previous AAgAtlas 1.0 database, which constitutes the largest human AAg database to date (Figure A,B, Figure S9). A Venn diagram analysis shows that our database has 6919 more AAgs than our previously developed AAgAtlas 1.0 database () and has 3575 more AAgs than the AAgMarker 1.0 database (Figure C). The AAgs are associated with 1073 human diseases (Figure A), including cancer and diseases related to the immune, nervous, cardiovascular, and gastrointestinal systems (Table S1).…”

“…It is also important to mention that a high number of AAbs (n = 4470) was compiled in 2018 by Zhu et al using a different text mining method to generate the AAgMarker database. 17 Further studies are warranted to understand why so many proteins have the potential to be an AAg. 60…”

“…14 Since the first report of autoantibodies in 1945, 15 hundreds of AAbs have been identified for a variety of different human diseases. 16,17 However, the information about the immunogenic properties of the AAgs remain limited. 18 In this work, we curated the largest human AAg database to date by searching human AAgs from PubMed abstracts, fulltext articles, and microarray datasets by text-mining and manual curation.…”

“…We also previously identified and validated five AAb biomarkers (PD1, P53, SIX2, EIF4E2, PD-L1) associated with the prognosis of patients of alveolar soft part sarcoma, non-small-cell lung cancer, and lymphoma treated with PD1 immunotherapy . Since the first report of autoantibodies in 1945, hundreds of AAbs have been identified for a variety of different human diseases. , However, the information about the immunogenic properties of the AAgs remain limited. − …”

Applications of Protein Microarrays in Biomarker Discovery for Autoimmune Diseases

“…Computational algorithm analysis could facilitate selection of appropriate biomarker profiles with better predictive accuracy ( 30 ). For data sharing, online databases such as Protein Microarray Database ( 130 ), AAgMarker 1.0 ( 131 ), and AAgAtlas 1.0 ( 132 ) offer a comprehensive platform to overview data from protein microarrays as well as various candidate autoantibody biomarkers in different diseases. To validate the identified novel biomarkers, traditional methods such as ELISA and immunoblotting would be preferred because of their higher compatibility with the current popular methods in lab tests in hospitals.…”