Design and Implementation of a Scalable Python Package for Synthetic Data Augmentation and Missing Value Imputation in Large-Scale Proteomics Datasets

November 13, 2025

Missing Data is a long-standing problem across several fields, including proteomics, where data can be missing from protein and peptide measurements. In fact, this is something that happens in 5-50\% of the cases. Traditionally, missing values have been addressed using basic methods such as mean-imputation or machine-learning models. Recently, new deep-learning models are being applied to improve the process of dealing with missing values, particularly in complex proteomics datasets. This paper explores the process of building a functional Python package for an imputation model in the field of proteomics, allowing researchers to mitigate missing data. By complementing the model with tests, online documentation and practical examples, the package offers a complete tool that is both user-friendly and efficient. Additionally, it can be integrated with well-established proteomics systems, aiming to make missing data imputation more accurate and streamlined. Several metrics were used to validate the quality of the work, with the results showing a modular and usable final result. In conclusion, this work provides a complete tool ready to be used by any researcher working in proteomics. For future work, additional imputation models should be included in the package to increase the options available to users.

Authors in the community:

• 

  D

  Diogo Clemente dos Santos Tavares Ferreira

  ist199200

Supervisors of this institution:

• 

  E

  Emanuel José Vieira Gonçalves

  ist429071

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

eng - English

Instituto Superior Técnico