Distinguishing protein-coding from non-coding RNAs through support vector machines
2006

Distinguishing Protein-Coding from Non-Coding RNAs

Sample size: 102801 publication Evidence: high

Author Information

Author(s): Liu Jinfeng, Gough Julian, Rost Burkhard

Primary Institution: Columbia University

Hypothesis

The hypothesis is that native proteins have particular structural and sequence properties that distinguish them from putative translation products of non-coding RNAs.

Conclusion

The study developed a method that reliably distinguishes protein-coding RNAs from non-coding RNAs, identifying over 14,000 non-coding RNAs in mouse and estimating the total number to be about 28,000.

Supporting Evidence

  • The method achieved about 97% specificity and 98% sensitivity in distinguishing coding RNAs from non-coding RNAs.
  • CONC was applied to 102,801 mouse cDNAs from the FANTOM3 dataset.
  • The study identified over 14,000 non-coding RNAs in mouse.
  • Estimates suggest the total number of non-coding RNAs in the FANTOM3 dataset is about 28,000.

Takeaway

This study created a computer program that can tell the difference between two types of RNA: one that makes proteins and one that doesn't, helping scientists understand more about genes.

Methodology

The study used support vector machines (SVMs) to classify RNA transcripts based on various protein-related features.

Potential Biases

The method may be biased towards known proteins, potentially misclassifying novel transcripts.

Limitations

The performance of the method may vary based on the dataset and the specific characteristics of the RNA being analyzed.

Statistical Information

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1371/journal.pgen.0020029

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication