Back

RBCeq: An Integrated Bioinformatics Algorithm Designed to Improve Blood Type Compatibility Testing

Jadhao, S. S.; Davison, C.; Roulis, E. V.; Schoeman, E. M.; Divate, M. D.; Shankar, A. J.; Lee, S.; Pecheniuk, N. M.; Irving, D. O.; Hyland, C. A.; Flower, R. L.; Nagaraj, S. H.

2021-01-15 bioinformatics
10.1101/2021.01.13.426510 bioRxiv
Show abstract

While blood transfusion is an essential cornerstone of hematological care, patients that require repetitive transfusion remain at persistent risk of alloimmunization due to the diversity of human blood group polymorphisms. Next-generation sequencing (NGS) is an effective means of identifying genotypic and phenotypic variations among the blood groups, while the accurate interpretation of such NGS data is currently hampered by a lack of accessibility to bioinformatics support. To address this unmet need, we have developed the RBCeq (https://www.rbceq.org/) platform, which consists of a novel bioinformatics algorithm coupled with a user-friendly web server capable of comprehensively delineating different blood group variants from genomics data with advanced visualization of results. The software profiles genomic data for 36 blood group systems, including two transcription factors and can identify small genetic alterations, including small indels and copy number variants. The RBCeq algorithm was validated on 403 samples which include 58 complex serology cases from Australian Red Cross LifeBlood, 100 samples from The MedSeq Project (phs000958) and a further 245 from Indigenous Australian participants. The final blood typing data from RBCeq was 99.83% concordant for 403 samples (85 different antigens in 21 blood group systems) with that listed from the International Society for Blood Transfusion database.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.