SEQENS: An ensemble method for relevant gene identification in microarray data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate identification of relevant variables is crucial in case-control studies, particularly with high-dimensional genetic expression microarray data.
• Existing feature selection and identification methods may lack the stability or accuracy required for complex biological datasets.

Purpose of the Study:

• To introduce and evaluate SEQENS, an ensemble feature identification algorithm designed for case-control studies.
• To assess SEQENS's performance in identifying relevant genes using genetic expression microarray data and synthetic groundtruths.
• To present Cliff, a novel metric for evaluating feature identification algorithm performance when true relevant variables are known.

Main Methods:

• SEQENS employs Sequential Feature Search with multiple sample splitting to identify variables with a strong relationship to the target.
• The algorithm generates a variable relevance ranking, applicable for both feature identification and selection.
• Performance was validated using the E-MTAB-3732 dataset with generated synthetic groundtruths and varying sample-to-dimensionality ratios.

Main Results:

• SEQENS demonstrated superior performance in identifying relevant genes compared to state-of-the-art methods on average.
• The algorithm exhibited enhanced stability in its identification of relevant variables.
• The Cliff metric provided a robust evaluation of algorithm performance against known groundtruths.

Conclusions:

• SEQENS is a robust and stable algorithm for feature identification in genetic expression microarray data.
• The algorithm offers a valuable tool for researchers in case-control studies, potentially improving classifier optimization.
• SEQENS is publicly available, promoting its adoption and further development within the research community.