Improving Predictions of Mitochondrial Protein Localization
Author Information
Author(s): Kieren T. Lythgow, Gavin Hudson, Peter Andras, Patrick F. Chinnery
Primary Institution: Newcastle University
Hypothesis
Increasing the number of independent prediction methods can paradoxically decrease the accuracy of predicting mitochondrial localization.
Conclusion
Using a combination of seven carefully selected classifiers yields the best sensitivity and specificity for predicting mitochondrial localization.
Supporting Evidence
- Over 50% of OXPHOS disorders are present in childhood.
- Using a larger group of prediction tools would compromise the approach.
- The highest sensitivity was obtained when four specific parameters were excluded.
- Mean sensitivity using the best combination of seven prediction tools was significantly greater than using all 11 tools.
Takeaway
This study shows that using too many methods to predict where proteins go in cells can actually make it harder to get it right, but using the right seven methods can help a lot.
Methodology
The study evaluated the performance of different combinations of bioinformatic prediction tools using known mitochondrial and non-mitochondrial proteins.
Potential Biases
The inclusion of data sets from lower organisms may restrict the sensitivity of detection without increasing specificity.
Limitations
The study's predictions may vary considerably each time due to the inherent variability in the sensitivity of the prediction methods.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website