Although the prediction program used by DOOR was ranked as the most accurate among 14 currently popular operon prediction programs [11], [12], it still generated false operon structures.
HGT candidates were confirmed to be KNOTTIN proteins using the prediction program provided by the KNOTTIN database [ 31, 69].
Even more important, false positives might be mainly caused by limits of the prediction programs.
We investigated the possible reason for differences in the TP rates observed for the Arabidopsis and non-Arabidopsis species by considering (a) commonality in the prediction programs, and (b) characterizing features of interactions for the miRNAs that were falsely predicted to be negative.
All mutations were predicted as not tolerated by the SIFT prediction program and predicted as probably damaging by PolyPhen2, except for p.Ala284Asp (c.851C > A), which was predicted as possibly damaging.
One objective was to determine if AtAPY1 was a soluble protein in the lumen of the Golgi or, as implied by the TMHMM prediction program [ 69], a Golgi membrane protein with an uncleaved signal sequence at the N-terminus serving as a transmembrane anchor.
Among the putative sRNA genes detected in GAS M49 by the tiling array approach, two sequences were categorised by the Rfam prediction program as CRISPR-related RNAs (Table 1).
It is also worth noting that the missing pieces each correspond to one or more contiguous exons; in other words the differences between the chimp model and the human protein are entirely explainable by the gene prediction program skipping exons.
Note that the JGI genes were predicted by the gene prediction programs Genscan, Fgenesh and Genewise, and have not been subjected to rigorous annotation criteria.
One of the reasons why these genes were not detected by the gene prediction programs could be the presence of low quality sequences that may have lead to frameshifts, thus making them impossible to detect by these programs.
