SEL.TS.AREA

SEL.TS.AREA identifies differentially expressed genes from microarray time-series expression data under data-poor conditions to detect temporal changes in transcript abundance.

Key Features:

• Implementation: An R script that implements three analytical methods for time-series differential expression detection.
• Data-poor context adaptation: Evaluates differential expression with limited replicates specifically for small-sample microarray time series.
• Method 2 — Area-based selection: Computes the area bounded by time-series expression profiles and calls genes differentially expressed when the area exceeds a threshold derived from an experimental error model.
• Method 1 — Sample-thresholding: Applies thresholds on individual samples based on experimental error models.
• Method 3 — Spline comparison: Uses splines to compare time-series profiles across conditions.
• Comparative performance: Synthetic-data benchmarking reported superior Precision and Recall for Method 2 on short time series and better performance of Method 3 for longer series.
• Application to real data: Methods have been applied to real datasets for empirical validation.

Scientific Applications:

• Microarray time-series studies: Selection of differentially expressed transcripts in experiments with limited biological replicates.
• Precision and recall optimization: Choice of method to optimize Precision and Recall according to time-series length (Method 2 for short, Method 3 for long).
• Method selection guidance: Comparative results from synthetic and real datasets inform selection among the three algorithms for specific experimental conditions.

Methodology:

Evaluates three explicit methods: Method 1 applies thresholds on individual samples using experimental error models; Method 2 computes the area bounded by time-series expression profiles and compares it to a threshold from an experimental error model; Method 3 compares profiles using splines; comparative assessment used synthetic data and applications to real datasets.