Detailed estimation of bioinformatics prediction reliability through the Fragmented Prediction Performance Plots
2007

Estimating Bioinformatics Prediction Reliability

publication Evidence: moderate

Author Information

Author(s): Oliviero Carugo

Primary Institution: Pavia University and Max F. Perutz Laboratories, Vienna University

Hypothesis

The reliability of bioinformatics predictions depends on the amount of data used for predictions.

Conclusion

Using Fragmented Prediction Performance Plots (FPPPs) is essential to quantify the performance of bioinformatics prediction protocols.

Supporting Evidence

  • The reliability of predictions improves with larger datasets.
  • FPPPs can help determine if a prediction method has reached its best performance.
  • Different prediction methods may behave differently based on the amount of data.

Takeaway

This study shows that the more data you have, the better your predictions can be, but you need to check if adding more data will actually help.

Methodology

The study describes a tool called Fragmented Prediction Performance Plot (FPPP) to analyze the relationship between prediction reliability and data amount.

Potential Biases

The predictions may be influenced by the specific characteristics of the training sets used.

Limitations

The analysis may not account for all factors affecting prediction quality, such as data noise and heterogeneity.

Digital Object Identifier (DOI)

10.1186/1471-2105-8-380

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication