Humanness Evaluation
1 Introduction
The module can determine the probability that an antibody belongs to human based on its V-region sequence.
The model is trained based on 140 million antibody V-region sequences from Observed Antibody Space database (OAS)1,2, and has an accuracy of 99% in predicting whether an antibody species belongs to humans. In the ADA data disclosed by FDA, the model can significantly identify fully human antibodies, and the predicted scores are negatively correlated with ADA, further proving the reliability of the model.
2 Performance
Figure 1. Confusion matrix of the model on test set.
Figure 2. Reliability curve of the model on test set.
Figure 3. Model predicted humanness scores are negatively correlated with clinical ADA disclosed by FDA.
Figure 4. The tool accurately predicts the humanness of anti-PD-L1 antibodies. Among them, C57BL/6 refers to antibodies produced by wild-type mice, while HUGO-Ab refers to antibodies produced by HUGOTM fully human antibody mice.
3 Parameters
- Fv Sequence: Amino acid sequences of antibody Fv regions in FASTA format. IIf the input sequences contain parts outside the V regions, the tool will automatically identify and extract the V regions.
4 Results Explanation
The numerical Score takes values in the range [0,1], the closer to 1 the higher the probability of being human antibodies.
5 Reference
[1] Olsen TH, Boyles F, Deane CM. Observed Antibody Space: A diverse database of cleaned, annotated, and translated unpaired and paired antibody sequences. Protein Sci. 2022 Jan;31(1):141-146. doi: 10.1002/pro.4205. Epub 2021 Oct 29. PMID: 34655133; PMCID: PMC8740823. https://doi.org/10.1002/pro.4205
[2] Aleksandr Kovaltsuk, Jinwoo Leem, Sebastian Kelm, James Snowden, Charlotte M. Deane, Konrad Krawczyk; Observed Antibody Space: A Resource for Data Mining Next-Generation Sequencing of Antibody Repertoires. J Immunol 15 October 2018; 201 (8): 2502–2509. https://doi.org/10.4049/jimmunol.1800708