Introduction: The objective of this study is to evaluate a novel
gradient boosting methodology on established metabolic pathway
prediction benchmarks to solve the problem of reconstructing
metabolic networks of organisms from genome sequences. Metabol...
Introduction: The objective of this study is to evaluate a novel
gradient boosting methodology on established metabolic pathway
prediction benchmarks to solve the problem of reconstructing
metabolic networks of organisms from genome sequences. Metabolic
pathway prediction is one approach to address the problem in system
biology of reconstructing the metabolic network of an organism from
its genome sequence. Recently there have been developments in
machine learning-based pathway prediction methods that conclude
that machine learning-based approaches are similar in performance
to the most used method, PathoLogic which is a rule-based method.
One issue is that previous studies evaluated PathoLogic without
taxonomic pruning which decreases its performance.
Results: This study updates the evaluation results from previous
studies to demonstrate that PathoLogic with taxonomic pruning
outperforms previous machine learning-based approaches and that
further improvements in performance need to be made for them to be
competitive. Furthermore, this study introduces mlGBPR, a gradient
boosting-based metabolic pathway prediction method improving on
the multi-label classification pathway prediction framework
introduced from mlLGPR. An improvement to this multi-label
framework is made by utilizing correlations between labels using
classifier chains. A ranking method is proposed that determines the
order of the chain so that lower performing classifiers are placed
later in the chain to utilize the correlations between labels more.
Finally, mlGBPR is evaluated with and without classifier chains on
single-organism and multi-organism benchmarks. Results indicate
that mlGBPR outperforms other previous pathway prediction methodsii
including PathoLogic with taxonomic pruning in terms of hamming
loss, precision and F1 score on single organism benchmarks.
Discussion: The results from this study indicate that the performance
of machine learning-based pathway prediction methods can be
substantially improved and can even outperform PathoLogic with
taxonomic pruning. The methods developed in this study can
generate higher quality predictions that can be used to expediate the
processing of creating metabolic pathway models. These models
have further applications in medicine to construct integrated human-
virus metabolic models where candidate drug-gene targets for
diseases such as COVID-19 can be identified through computational
analysis that minimize viral metabolism and host damage.