In an attempt to overcome the bias from different annotations all genomes were annotated using the Prodigal gene finder [ 23] which provided consistency across the entire data set.
In only a few cases were there conspicuous differences between the two different annotations (fig. 3).
Also, differences were observed when comparing different annotations of the same genome assembly (NCBI-UMD3.1 vs. Ensembl-UMD3.1).
The example above can be sorted out by a human reading the slightly different annotations, but additional examples of differences in annotation are more difficult to reconcile.
To better understand the discrepancies between annotators, we asked them to review the different annotations and reach a consensus.
For the interaction probabilities, N is the number of interactions in our network between proteins with at least two different annotations, K is the number of interactions involving proteins annotated to GO1, n the number of interactions involving proteins annotated to GO2 and k the the number of interactions between a protein annotated to GO1 and one annotated to GO2.
Then, we calculated a Bray Curtis dissimilarity matrix between samples for the three different annotations to compare them.
One is to diversify your data set and to have many different annotations of your data set, like race and gender and age.
As Fig. 4 shows, questions with different annotations have clearly different depth score means.
Co-workers could then discuss comments in-line, vote up and down different annotations, and work towards a finished document.
To see this, suppose we have three different annotations for an object, b i, b i+1, and b i+2, as in Fig. 10.
