Author name cluster

Tianxiao Li

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

5 papers

2 author rows

AAAI Conference 2025 Conference Paper

Learning Disentangled Equivariant Representation for Explicitly Controllable 3D Molecule Generation

Haoran Liu
Youzhi Luo
Tianxiao Li
James Caverlee
Martin Renqiang Min

We consider the conditional generation of 3D drug-like molecules with explicit control over molecular properties such as drug-like properties (e.g., Quantitative Estimate of Druglikeness or Synthetic Accessibility score) and effectively binding to specific protein sites. To tackle this problem, we propose an E(3)-equivariant Wasserstein autoencoder and factorize the latent space of our generative model into two disentangled aspects: molecular properties and the remaining structural context of 3D molecules. Our model ensures explicit control over these molecular attributes while maintaining equivariance of coordinate representation and invariance of data likelihood. Furthermore, we introduce a novel alignment-based coordinate loss to adapt equivariant networks for auto-regressive de-novo 3D molecule generation from scratch. Extensive experiments validate our model's effectiveness on property-guided and context-guided molecule generation, both for de-novo 3D molecule design and structure-based drug discovery against protein targets.

PDF Details DOI

SODA Conference 2024 Conference Paper

Dynamic Dictionary with Subconstant Wasted Bits per Key

Tianxiao Li
Jingxun Liang
Huacheng Yu
Renfei Zhou

Dictionaries have been one of the central questions in data structures. A dictionary data structure maintains a set of key-value pairs under insertions and deletions such that given a query key, the data structure efficiently returns its value. The state-of-the-art dictionaries [4] store n key-value pairs with only O(n log (k) n) bits of redundancy, and support all operations in O ( k ) time, for k ≤ log* n. It was recently shown to be optimal [16]. In this paper, we study the regime where the number of redundant bits is R = o ( n ), and show that when R is at least n /poly log n, all operations can be supported in O (log* n + log( n / R )) time, matching the lower bound in this regime [16]. We present two data structures based on which range R is in. The data structure for R < n /log 0. 1 n utilizes a generalization of adapters studied in [5, 15]. The data structure for R ≥ n /log 0. 1 n is based on recursively hashing into buckets with logarithmic sizes.

Details

NeurIPS Conference 2023 Conference Paper

Disentangled Wasserstein Autoencoder for T-Cell Receptor Engineering

Tianxiao Li
Hongyu Guo
Filippo Grazioli
Mark Gerstein
Martin Renqiang Min

In protein biophysics, the separation between the functionally important residues (forming the active site or binding surface) and those that create the overall structure (the fold) is a well-established and fundamental concept. Identifying and modifying those functional sites is critical for protein engineering but computationally non-trivial, and requires significant domain knowledge. To automate this process from a data-driven perspective, we propose a disentangled Wasserstein autoencoder with an auxiliary classifier, which isolates the function-related patterns from the rest with theoretical guarantees. This enables one-pass protein sequence editing and improves the understanding of the resulting sequences and editing actions involved. To demonstrate its effectiveness, we apply it to T-cell receptors (TCRs), a well-studied structure-function case. We show that our method can be used to alter the function of TCRs without changing the structural backbone, outperforming several competing methods in generation quality and efficiency, and requiring only 10\% of the running time needed by baseline models. To our knowledge, this is the first approach that utilizes disentangled representations for TCR engineering.

PDF Details

FOCS Conference 2023 Conference Paper

Dynamic "Succincter"

Tianxiao Li
Jingxun Liang
Huacheng Yu
Renfei Zhou

Augmented B-trees (aB-trees) are a broad class of data structures. The seminal work “succincter” by Pǎtraşcu [1] showed that any aB-tree can be stored using only two bits of redundancy, while supporting queries to the tree in time proportional to its depth. It has been a versatile building block for constructing succinct data structures, including rank/select data structures, dictionaries, locally decodable arithmetic coding, storing balanced parenthesis, etc. In this paper, we show how to “dynamize” an aB-tree. Our main result is the design of dynamic aB-trees (daB-trees) with branching factor two using only three bits of redundancy (with the help of lookup tables that are of negligible size in applications), while supporting updates and queries in time polynomial in its depth. As an application, we present a dynamic rank/select data structure for n-bit arrays, also known as a dynamic fully indexable dictionary (FID) [2]. It supports updates and queries in $O(\log n / \log \log n)$ time, and when the array has m ones, the \begin{equation*}\log \begin{pmatrix}n \\m\end{pmatrix}+On / 2^{\log 0. 199} n\end{equation*}bits. Note that the update and query times are optimal even without space constraints due to a lower bound by Fredman and Saks [3]. Prior to our work, no dynamic FID with near-optimal update and query times and redundancy $o(n / \log n)$ was known. We further show that a dynamic sequence supporting insertions, deletions and rank/select queries can be maintained in (optimal) $O(\log n / \log \log n)$ time and with $O\left(n \cdot \operatorname{poly} \log \log n / \log ^{2} n\right)$ bits of redundancy.

Details

FOCS Conference 2023 Conference Paper

Tight Cell-Probe Lower Bounds for Dynamic Succinct Dictionaries

Tianxiao Li
Jingxun Liang
Huacheng Yu
Renfei Zhou

A dictionary data structure maintains a set of at most n keys from the universe $[U]$ under key insertions and deletions, such that given a query $x \in[U]$, it returns if x is in the set. Some variants also store values associated to the keys such that given a query x, the value associated to x is returned when x is in the set. This fundamental data structure problem has been studied for six decades since the introduction of hash tables in 1953. A hash table occupies $O(n \log U)$ bits of space with constant time per operation in expectation. There has been a vast literature on improving its time and space usage. The state-of-the-art dictionary by Bender, Farach-Colton, Kuszmaul, Kuszmaul and Liu [1] has space consumption close to the information-theoretic optimum, using a total of \begin{equation*}\log \begin{pmatrix} U \\ n \end{pmatrix}+On\log ^{\left(k\right)} n\end{equation*} bits, while supporting all operations in $O(k)$ time, for any parameter $k \leq \log ^{*} n$. The term $O\left(\log ^{(k)} n\right)=O(\underbrace{\log \cdots \log n})$ is referred to as the wasted bits per key. In this paper, we prove a matching cell-probe lower bound: For $U=n^{1+\Theta(1)}$, any dictionary with $O\left(\log ^{(k)} n\right)$ wasted bits per key must have expected operational time $\Omega(k)$, in the cell-probe model with word-size $w=\Theta(\log U)$. Furthermore, if a dictionary stores values of $\Theta(\log U)$ bits, we show that regardless of the query time, it must have $\Omega(k)$ expected update time. It is worth noting that this is the first cell-probe lower bound on the trade-off between space and update time for general data structures.

Details