Philip M. Kim

Neopep: Reinforcement Learning From Machine-Learned Force Field Energy for Non-Canonical Peptide Structure Prediction

Neopep uses reinforcement learning with machine-learned force field energy to predict structures of non-canonical amino acid-containing peptides, achieving lower RMSD than existing methods.

Design of peptides with non-canonical amino acids using flow matching

This paper introduces NCFlow, a generative model for designing peptides with non-canonical amino acids, showing improved binding affinity over existing methods.

ZATELLITE: a toolkit to visualize and manipulate human centromeres in live cells using synthetic zinc fingers

ZATELLITE enables real-time visualization and manipulation of human centromeres in live cells using synthetic zinc fingers without causing DNA damage.

SLiMNet: a deep learning model to detect short linear motifs using protein large language model representations and paired inputs

SLiMNet detects functional similarity in SLiM pairs using deep learning and protein LLM embeddings, identifying new motifs and providing functional annotations.

Making invisible excited state protein structures visible by combining NMR and machine learning

The paper combines NMR data with AlphaFlow to identify excited state conformers of pro-IL-18, revealing structural insights from transient states.

Democratizing Protein Language Model Training, Sharing and Collaboration

The paper introduces SaprotHub, a platform that simplifies training and sharing of protein language models, enabling non-experts to build and collaborate on customized models.

CyclicBoltz1, fast and accurately predicting structures of cyclic peptides and complexes containing non-canonical amino acids using AlphaFold 3 Framework

The paper extends the AlphaFold3 framework to predict structures of cyclic peptides with non-canonical amino acids, outperforming existing models on a test set.

FlowPacker: Protein side-chain packing with torsional flow matching

FlowPacker predicts protein side-chain conformations using torsional flow matching and equivariant graph attention, outperforming existing methods and handling missing coordinates and multimeric targets.

EuDockScore: euclidean graph neural networks for scoring protein-protein interfaces

This paper introduces EuDockScore, a scoring model for protein-protein interfaces using euclidean graph neural networks, and provides variants for antibody-antigen docking and re-ranking AlphaFold-Multimer outputs.

PepFlow: direct conformational sampling from peptide energy landscapes through hypernetwork-conditioned diffusion

PepFlow samples peptide conformations using a diffusion-based Boltzmann generator with hypernetworks, achieving faster and accurate predictions than traditional methods.

ProteinSGM: Score-based generative modeling for de novo protein design

This paper introduces ProteinSGM, a score-based diffusion model for de novo protein design, showing it can generate native-like structures and perform conditional design.

GDockScore: a graph-based protein-protein dockingscoring function

GDockScore is a graph-based deep learning model for protein-protein docking scoring, achieving state-of-the-art performance on challenging datasets.

PepNN: a deep attention model for the identification of peptide binding sites

PepNN introduces deep attention models for predicting peptide binding sites on proteins using sequence and structural data, achieving high performance on benchmark tests.

Fast and flexible design of novel proteins using graph neural networks

A deep graph neural network trained on Sudoku puzzles designs novel protein sequences that adopt predetermined structures.

A multi-reporter bacterial 2-hybrid assay for the fast, simple, and dynamic assay of PDZ domain - peptide interactions

The paper introduces a bacterial 2-hybrid assay for screening PDZ domain-peptide interactions with improved sensitivity and scalability.