Systematic contextual biases in SegmentNT potentially relevant to other nucleotide transformer models
Mark Ebbert, Anna Ho, Madeline L. Page, Bram Dutch, Blake K Byer, Kristen L. Hankins, Hady W. Sabra, Bernardo Aguzzoli Heberle, Mark E. Wadsworth, Grant A Fox, Bikram Karki, Caylin D. Hickey, David W. Fardo, Cody Bumgardner, Yasminka A. Jakubek, , Justin B. Miller
The paper identifies contextual biases in the SegmentNT model, showing how input sequence length and nucleotide position affect predictions and suggests ways to mitigate these biases.
Nucleic Acids Research · Jul 17, 2026Peer-reviewedBioinformatics