Examples of Consistency Checking and Reconciliation

Inconsistencies can arise from a variety of sources. Some examples include the following:

  1. Misspellings (typos). One might, for example, incorrectly enter the name of a transcription factor. These are difficult to distinguish from similarly named transcription factors that are actually different. As a result they are handled by printing warnings which the experimenter must check manually. An example of this situation is shown in Misspelling Example. In this example, the HSP40 was incorrectly written as HSP400 at one place. Because HSP400 is not the name of a gene, the data is inconsistent. Because there is a valid gene name that has a similar name, the consistency checker would produce an error message stating that the most likely correct name is HSP400.
  2. Motifs generated from the same data set by different algorithms. The algorithms use very different assumptions, so they can produce markedly different results. An example of this is shown in Example of Fusion of AlignAce and BioProspector. In this case, the output of the two algorithms is fused by averaging.
  3. Motifs generated from different data sets or by different runs on the same data set by the same algorithm. An example of this is shown in Example of Fusion of Different Runs of Gibbs Motif Sampler. In the example, 3 runs were fused.