- Hu-mAb is an antibody humanisation tool.
- Using large-scale sequence data from OAS, we generated Random Forest models that classify antibody variable domain sequences as human/non-human.
- By making mutations that increase the 'humanness' score, we can efficiently humanise an antibody sequence, making it less likely to be immunogenic.
- Mutations are only made to framework regions; CDR residues are left alone to maintain the antibody binding properties.
- If not specified, the V-gene family to which your sequence will be compared is selected by evaluating the humanness score for the sequence compared to each V-gene type, choosing the highest-scoring.
- The Random Forests for each V-gene type have default 'threshold' scores - a humanness score above this value would mean that the sequence is classified as human, and so by default this is the score the humaniser will try to reach. However, if you would like to set your own threshold value you can.
- An example of the output produced by Hu-mAb can be seen here.
The paper describing Hu-mAb and its results can be found here:
Claire Marks, Alissa M Hummer, Mark Chin, Charlotte M Deane. Humanization of antibodies using a machine learning approach on large-scale repertoire data, Bioinformatics (2021).