Synthesizer: Expediting synthesis studies from context-free data with information retrieval techniques

Area of Science:

• Bioinformatics
• Data Science
• Computational Biology

Background:

• High-quality scientific datasets are increasingly available but often stored in unstructured spreadsheets.
• Lack of standardized annotations hinders data synthesis across different studies.
• Manual data merging is time-consuming, error-prone, and costly, creating a barrier to research.

Purpose of the Study:

• To develop an automated algorithm for merging unstructured spreadsheet data.
• To overcome the limitations of manual data curation in research synthesis.
• To improve the efficiency and accuracy of combining diverse datasets.

Main Methods:

• An information retrieval-inspired algorithm named Synthesize was developed.
• The algorithm merges unstructured data based on column labels and values.
• Synthesize was implemented in an open-source web application, Synthesizer, accepting CSV files.

Main Results:

• The Synthesize algorithm demonstrated high accuracy (85-100%) when applied to cancer and ecological datasets.
• The Synthesizer web application provides a user-friendly interface for data merging.
• The software visualizes merged data and outputs results as a new spreadsheet.

Conclusions:

• Automated data merging using Synthesize significantly reduces barriers to research synthesis.
• The Synthesizer tool offers an accurate and efficient solution for combining disparate datasets.
• Future work includes handling continuous data with unit detection and supporting database formats.