TY - JOUR
T1 - Stack-DCL
T2 - A Density-Aware and Local-Geometry Stacking Ensemble for Cell-Penetrating and Quorum-Sensing Peptide Classification
AU - Kang, Xiaorui
AU - Li, Jinjin
AU - Lam, Chan Tong
AU - Zou, Quan
AU - Lin, Changhang
AU - Wei, Leyi
N1 - Publisher Copyright:
© 2026 American Chemical Society
PY - 2026/10/16
Y1 - 2026/10/16
N2 - Cell-penetrating peptides (CPPs) and quorum-sensing peptides (QSPs) represent peptide functions related to intracellular delivery and microbial signaling. Sequence-based prediction remains difficult because benchmark embeddings often show uneven local support, overlapping class regions, and high dimensionality. A single classifier may therefore miss either local or global structure. We propose Stack-DCL, an out-of-fold stacking framework for CPP and QSP classification with fixed UniRep embeddings. Stack-DCL combines a density-aware weighted random forest, a cosine-neighborhood model, and an L2-regularized logistic regression classifier. The meta-learner is also logistic regression and fuses only positive-class probabilities generated from held-out folds, which reduces fusion-stage leakage. On benchmark train/test splits with five-fold cross-validation on training data and final held-out testing, Stack-DCL achieves independent-test AUC values of 0.980 for QSP and 0.974 for CPP. Ablation results indicate that the three branches capture local support, directional neighborhood similarity, and global linear separability as different representation-level signals. These results support Stack-DCL as a benchmark-level stacking design for the two evaluated peptide tasks.
AB - Cell-penetrating peptides (CPPs) and quorum-sensing peptides (QSPs) represent peptide functions related to intracellular delivery and microbial signaling. Sequence-based prediction remains difficult because benchmark embeddings often show uneven local support, overlapping class regions, and high dimensionality. A single classifier may therefore miss either local or global structure. We propose Stack-DCL, an out-of-fold stacking framework for CPP and QSP classification with fixed UniRep embeddings. Stack-DCL combines a density-aware weighted random forest, a cosine-neighborhood model, and an L2-regularized logistic regression classifier. The meta-learner is also logistic regression and fuses only positive-class probabilities generated from held-out folds, which reduces fusion-stage leakage. On benchmark train/test splits with five-fold cross-validation on training data and final held-out testing, Stack-DCL achieves independent-test AUC values of 0.980 for QSP and 0.974 for CPP. Ablation results indicate that the three branches capture local support, directional neighborhood similarity, and global linear separability as different representation-level signals. These results support Stack-DCL as a benchmark-level stacking design for the two evaluated peptide tasks.
KW - cell-penetrating peptides
KW - density-aware learning
KW - local geometry modeling
KW - quorum-sensing peptides
KW - stacking ensemble
KW - uniRep
UR - https://www.scopus.com/pages/publications/105053838594
U2 - 10.1021/acssynbio.6c00522
DO - 10.1021/acssynbio.6c00522
M3 - Article
C2 - 42916603
AN - SCOPUS:105053838594
SN - 2161-5063
VL - 15
SP - 4316
EP - 4326
JO - ACS Synthetic Biology
JF - ACS Synthetic Biology
IS - 10
ER -