Back

Sysrecon: A systematic data-driven tool for genome-scale metabolic reconstructions

Ouyang, S.; Li, Z.; Hu, J.; Cao, M.; Wang, X.; Sun, H.; Yu, F.; Mao, L.

2023-12-13 systems biology
10.1101/2023.12.13.571370 bioRxiv
Show abstract

Genome-scale metabolic reconstructions contain the information about metabolism, which understands organisms molecular mechanisms better. Different reconstruction procedures are developed to achieve high-quality GSMRs of organisms; however, the descriptions of procedures are vague, meaning that they cannot be qualitatively reproduced, even contradictory, and different published reconstructions may involve different steps. This indicates that there is no unified strategy to quantify the steps in GSMRs. To resolve the problem, we created a R package, Sysrecon, to quantitatively and qualitatively describe the steps of GSMRs in the literature. Sysrecon disassembles GSMR into 93 steps, which are compiled after the comprehensive textual analysis of metabolic reconstructions. Then, Sysrecon decomposes each step of the reconstruction procedure into three components: step content, step conversion, and database and tools. When building a GSMR, Sysrecon creates a template procedure that includes a list of pipeline steps, the description of the information conversion in the steps, and the database and tools that may be used to facilitate the conversion. Because each steps conversion is defined by a formula, the entire GSMR can be dismantled into many step blocks. The reconstruction procedure is therefore transformed into an adaptable automated assembly line that can be customized and tailored to settings of different organisms. To the best of our knowledge, Sysrecon is the first computational tool that dynamically constructs the GSMRs pipeline and provides a framework for defining reconstruction steps and can be used as a basis for the development of high-quality GSMRs.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.