Weighted persistent homology for biomolecular data analysis

Zhenyu Meng, D. Vijay Anand, Yunpeng Lu, Jie Wu, Kelin Xia

Abstract

In this paper, we systematically review weighted persistent homology (WPH) models and their applications in biomolecular data analysis. Essentially, the weight value, which reflects physical, chemical, and biological properties, can be assigned to vertices (atom centers), edges (bonds), or higher-order simplexes (clusters of atoms), depending on the biomolecular structure, function, and dynamics properties. Further, we propose the first localized weighted persistent homology (LWPH). Inspired by the success of element-specific persistent homology (ESPH), we do not treat biomolecules as an inseparable system like all previous weighted models; instead, we decompose them into a series of local domains, which may overlap. General persistent homology or weighted persistent homology analysis is then applied to each of these local domains. In this way, functional properties embedded in local structures can be revealed. Our model has been applied to systematically studying DNA structures, successfully discriminating between A-, B-, and Z-types of DNA. More importantly, our LWPH-based PCA model can identify two configurational states of DNA structure in an ionic liquid environment, which can otherwise be revealed only by complicated helical coordinate systems. The consistency with helical-coordinate models demonstrates that our model captures local structure variations so well that it is comparable with geometric models, while offering greater flexibility in characterizing domains of arbitrary sizes and shapes.

Introduction

Biomolecular data is expanding at an unprecedented rate, with millions of sequences in GenBank and over 150,000 structures in the Protein Data Bank. To extract meaningful insights from this complexity, researchers often employ Topological Data Analysis (TDA), specifically Persistent Homology (PH). PH is adept at identifying the "shape" of data—such as holes, loops, and voids—across multiple scales. However, traditional PH faces two significant challenges in biological applications: homogeneity and global bias. Standard models often treat all atoms as identical points, ignoring chemical properties like atom type or partial charge, and they typically analyze molecules as single, inseparable systems. This global approach can obscure the subtle, local structural changes that drive biological function.

To address these limitations, this research introduces Localized Weighted Persistent Homology (LWPH), a framework that makes TDA both physics-aware and spatially localized.

Methodology

Weighted Persistent Homology (WPH)

The foundation of this work lies in WPH, where physical and chemical context is integrated into the topological framework. Weights are assigned to vertices (atoms), edges (bonds), or higher-order structures to represent properties such as atomic radii or interaction potentials. This allows the model to distinguish between different chemical environments that would appear identical to a standard PH model.

Localized Weighted Persistent Homology (LWPH)

The core innovation is the decomposition of the molecular structure into local domains, typically represented as overlapping spheres. The process follows three main steps:

  1. Decomposition: The molecule is broken into a series of local domains.
  2. Local WPH Application: Within each domain, a weighted distance matrix is applied. For DNA analysis, the authors utilize a weight definition that sets the distance between atoms in the same residue to zero. This effectively forces the model to focus on non-covalent interactions between different residues, which are critical for structural stability.
  3. Fingerprinting: The topological features are visualized as "Betti curves" or Persistent Betti Numbers (PBNs), serving as unique fingerprints for local regions.

Experimental Results and Discussion

DNA Classification

The LWPH model was evaluated using 30 PDB files representing three major DNA configurations: A-DNA, B-DNA, and Z-DNA. While standard Local Persistent Homology (LPH) struggled to establish clear boundaries between these types, the weighted version (LWPH) produced distinct clusters in Principal Component Analysis (PCA) plots. This demonstrates that the inclusion of chemical weights is essential for accurate structural discrimination.

Identifying Configurational States in Ionic Liquids

The model was further tested on Molecular Dynamics (MD) trajectories of a 16-mer DNA duplex in two environments: standard water (WAT) and a restrictive ionic liquid (IL). Global PCA models based on raw atom coordinates or standard PH failed to cluster the DNA states in ionic liquids. In contrast, LWPH identified two distinct configurational states corresponding to different ion-DNA binding modes.

Furthermore, the model quantitatively captured the "confinement effect" of ionic liquids. The distribution area in PCA space—a proxy for conformational fluctuation—was significantly lower in ionic liquids. For instance, the distribution area for DNA in ionic liquid (IL2) was 296 units, compared to 554 units in water (Wat1), representing a nearly 50% reduction in structural fluctuation.

Comparison with Geometric Models

Traditional helical coordinate systems are the gold standard for DNA analysis but are mathematically complex and often limited to very local regions (one or two base pairs). LWPH provides accuracy comparable to these geometric models while offering significantly more flexibility. It can quantitatively characterize structural information in domains of arbitrary size and shape where traditional measurements fail.

Conclusion

LWPH represents a significant advancement in making TDA a viable tool for structural biology. By combining physical weights with localized analysis, the model captures fine-grained structural nuances that global models miss. Future work will involve extending this framework to proteins and larger biomolecular complexes. Given that LWPH features are "unit-free" and highly sensitive to local variations, they are ideal candidates for integration into machine learning pipelines for drug discovery and protein folding predictions.

Summary updated

Abstract

In this paper, we systematically review weighted persistent homology (WPH) models and their applications in biomolecular data analysis. Essentially, the weight value, which reflects physical, chemical and biological properties, can be assigned to vertices (atom centers), edges (bonds), or higher order simplexes (cluster of atoms), depending on the biomolecular structure, function, and dynamics properties. Further, we propose the first localized weighted persistent homology (LWPH). Inspired by the great success of element specific persistent homology (ESPH), we do not treat biomolecules as an inseparable system like all previous weighted models, instead we decompose them into a series of local domains, which may be overlapped with each other. The general persistent homology or weighted persistent homology analysis is then applied on each of these local domains. In this way, functional properties, that are embedded in local structures, can be revealed. Our model has been applied to systematically studying DNA structures. It has been found that our LWPH based features can be used to successfully discriminate the A-, B-, and Z-types of DNA. More importantly, our LWPH based PCA model can identify two configurational states of DNA structure in ion liquid environment, which can be revealed only by the complicated helical coordinate system. The great consistence with the helical-coordinate model demonstrates that our model captures local structure variations so well that it is comparable with geometric models. Moreover, geometric measurements are usually defined in very local regions. For instance, the helical-coordinate system is limited to one or two basepairs. However, our LWPH can quantitatively characterize structure information in local regions or domains with arbitrary sizes and shapes, where traditional geometrical measurements fail.
BibTeX
@inproceedings{meng2020weighted,
  author = {Zhenyu Meng and D. Vijay Anand and Yunpeng Lu and Jie Wu and Kelin Xia},
  title = {{Weighted persistent homology for biomolecular data analysis}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-019-55660-3},
}
Figure