New Cyrus Biotechnology Antibody Structure Prediction Software “NextGen” Outperforms Schrödinger and Other Software in Third-Party Test
Monday, August 10, 2020
Cyrus Biotechnology, Inc., announced it has developed new antibody structure prediction software, Cyrus NextGen, that is significantly more accurate than other leading methods according to a third-party test administered by BIOCAD. NextGen will allow the development of more effective and safe antibody biologic therapeutics by Cyrus and its partners across a range of disease indications.
Rosetta’s “hybridize” algorithm has been the leader in multiple assessments of protein structure prediction over the last decades (such as CASP and CAMEO), but it had not been applied to antibodies. Cyrus scientists, led by co-founder Yifan Song in collaboration with David La, have developed a specialized version of hybridize for antibody structure prediction. Internal tests showed superior performance over Schrödinger, CCG and other methods using the published “AMA-II” dataset.
To fully validate NextGen, Cyrus and BIOCAD collaborated to perform an independent test of NextGen, Schrödinger and other methods. BIOCAD’s rigorous performance metrics of model accuracy over the critical target-recognition loops (CDRs) and overall structure showed NextGen performing significantly better than the other methods on a 26-antibody benchmark set. More details are available at cyrusbio.com/news.
Cyrus CEO Dr. Lucas Nivón said, “Dr. Song saw an opportunity for a major advance in antibody modeling – the most common type of biologic drug. Now we are pleased to see the results of his insight fully realized – we hope NextGen will enable new cures for patients not served by current treatments.”
