Using the default cut-off of P = 0, the model had sensitivity of 74.1%, specificity of 88%, PPV of 87% and NPV of 75.9%.
Using the default cut-off of P = 0, the model had sensitivity of 58.8%, specificity of 96.3%, positive predictive value (PPV) of 76.9% and negative predictive value (NPV) of 91.8%.
Based on their availability in the protein database and their diversity, we selected 18 transposases (Table 1), covering at least 8 subfamilies, and used them as queries in a batch TBLASTN search performed on the WGS sequences of 20 Drosophila genomes [ 28] and Piano, Cherbas [ 29], using the default cut-off value.
Of the 9643 mouse samples in the compendium, 480 were labeled as TF+TG+ using the default cut-offs.
All protein sequences derived from the candidate PLD genes collected were examined using the domain analysis programs, PFAM http://pfam.sanger.ac.uk/ and SMART http://smart.embl-heidelberg.de/ with the default cut off parameters.
This algorithms uses Blast similarity score to find best reciprocal hits between complete genomes (we used the default cut-off value, i.e., 10−).
Using 2.0 as the default cut-off score of PRINCESS [ 37], we found 1,067 low-confidence PPIs (47.1%) and 1,199 high-confidence PPIs (52.9%).
The reliability of the alignments was examined using Guidance [ 43] and trimmed alignments using sites that had values above the default cut-off of 0.93 were generated.
The reliability of the alignments was examined using Guidance [ 58, 59] and trimmed alignments using sites that had values above the default cut-off of 0.93 were generated.
All protein sequences derived from the BLAST searches were examined using domain-analysis programs, including Pfam [ 35] and SMART [ 36], with the default cut-off parameters.
We therefore set the default cut-off score for SRST to 10, and use this cut-off for all subsequent analyses reported below.
