Tianang Leng

Making Every Experiment Count: Analogical Molecular Property Prediction

Tianang Leng^†, Joseph Lee^†, Haydn Jones, Cesar de la Fuente, Jacob Gardner, Mark Yatskar

We introduce a framework that predicts molecular properties by drawing on various experimental evidence from related molecules. A learned property-transfer model estimates whether a measurement on one molecule applies to another, while context optimization selects relevant, informative, and diverse evidence for an LLM to reason over. Across six molecular property prediction tasks, our framework outperforms the strongest ML baselines by 11.9 percentage points on literature-derived benchmarks and 11.2 points on TDC in mean macro-F1.

When Reasoning Meets Its Laws

Junyu Zhang^†, Yifan Sun^†, Tianang Leng^†, Jingyan Shen^†, Liu Ziyin, Paul Pu Liang, Huan Zhang

The paper introduces LORE, a framework that defines how large reasoning models should scale their thinking and accuracy with problem complexity. It builds a benchmark (LORE-BENCH) to test two key properties—monotonicity and compositionality—and finds current models are mostly monotonic but not compositional; fine-tuning to enforce the compute rule improves overall reasoning performance.

Machine learning for biomaterials design

Xiaoqiong Xia^†, Tianang Leng^†, Andrés J. García, Cesar de la Fuente-Nunez

This review surveys how machine learning and deep learning accelerate biomaterials design across metals, polymers, ceramics, and biological materials. It connects property prediction, generative design, and active learning with public datasets and experimental workflows, and examines challenges in data quality, interpretability, and multi-scale modelling. We discuss how self-driving laboratories, multimodal data, and physics-informed models could support more autonomous and personalized biomaterials discovery.

Predicting and generating antibiotics against future pathogens with ApexOracle

Tianang Leng^†, Fangping Wan^†, Marcelo Der Torossian Torres, Cesar de la Fuente-Nunez

ApexOracle centers on a discrete diffusion language model (DLM) that does two jobs at once: it learns chemistry-aware features of molecules and it can run the diffusion process backward to generate new antimicrobial candidates. This DLM backbone is then plugged into a multimodal architecture that also ingests pathogen genomics (Evo2) and pathogen text knowledge (Me-LLaMA), and cross-attends these with the molecule features. That multimodal fusion is what lets ApexOracle stay effective on unseen or future strains: given a new genome or description, it can both predict activity and guide the DLM to sample a molecule that should work for that specific pathogen.

AI in biomaterials discovery: generating self-assembling peptides with resource-efficient deep learning

Tianang Leng, Cesar de la Fuente-Nunez

Here we argue that resource-efficient hybrid RNN deep-learning models can generatively design self-assembling peptides with strong self-organization, underscoring AI's expanding role in biomaterials discovery.

The gut's hidden arsenal: A genomics-guided atlas of class II bacteriocins

Tianang Leng, Cesar de la Fuente-Nunez

This study introduces a genomics-guided pipeline to uncover a vast reservoir of class II bacteriocins from the human gut microbiome. These unmodified peptides demonstrate potent, narrow-spectrum antimicrobial activity with minimal disruption to the commensal microbiota, offering a promising avenue for precision microbiome engineering and combating antimicrobial resistance.

Self-sampling meta SAM: enhancing few-shot medical image segmentation with meta-learning

Tianang Leng^†, Yiming Zhang^†, Kun Han, Xiaohui Xie

We introduce three novel modules for foundational vision models for efficient few-shot medical image segmentation, experiments on popular abdominal CT and MRI dataset showed average improvements of 10.21% and 1.80% in terms of DSC, respectively.

PepLink

Peptide to SMILES is not the hard part anymore. PepLink goes further: it supports non-canonical amino acids, diverse cyclic peptide structures, and can even reverse-parse cyclic peptide SMILES back into amino acid sequences. Built for real peptide workflows, not just toy examples.

Getting started is easy: pip install PepLink — or try it instantly in the browser via the Web UI.

Kaggle: Predict Student Performance from Game Play

Utilizing insights from BERT and MAE, we developed a sequence classification model to accurately predict students' performance through a series of gameplay actions, earning us a prestigious bronze medal (top 5.6%).

2022 World Robot Contest — BCI Controlled Robot Contest

Utilizing Euclidean Alignment and Independent Component Analysis (ICA), we generated a substantial volume of high-quality data, successfully mitigating the over-fitting issue prevalent in EEG based Brain-Computer Interfaces. Impressively, I secured the 20th position among 283 global teams predominantly comprised of master's and Ph.D. students.

2022 HUST-Cambridge Foundations of Data Science — Masked Face Recognition

By incorporating a lightweight Convolutional Block Attention Module into ArcFace and optimizing the training with challenging images isolated using MCTNN, I successfully honed the network's focus on distinguishing features such as the eyes and hairline, thereby enhancing the performance by a notable 4%.

Undergraduate Innovation & Entrepreneurship Project — Intelligent football game analysis

By adeptly merging a transformer-based tracker with YOLOv5, I successfully developed a system capable of mitigating player and ball occlusion issues, enabling automatic and precise tracking of the small soccer ball at real-time FPS through a whole game.