Human subject testing of alignments

[Bruce Kirchoff]

Dave and I were wondering if it would be a good idea to include some
reliability testing of alignments performed with DNA-align as part of
the grant proposal we are considering. Here is how Dave outlined the
idea this morning. A number of scientists would be given a sequence
alignment had been created with one of the automated tools. They would
bring it into DNA-align and complete the alignment using the tools
available in the program. We would collect the alignments and compare
them to see if people were able to agree on the final alignment, and/
or how similar the alignments were.

In order to do this we would need some way of automatically comparing
the alignments. The GLOSA scores (or graphs) would give us one way of
doing this, but we would also need to know exactly where the
alignments did not agree. So there would have to be some way of doing
pairwise comparisons between the alignments and highlighting the
differences. Would this be possible? Would this add anything to the
grant?

I have worked with human subjects, and could lead that part of the
project. However, everyone involved in handling the data would have to
complete the human subjects training at your respective universities,
and the protocol for the experiment would have to be cleared by all
the universities where investigators were involved. There are ways
that we can minimize the amount of work that is needed in this
process, but you should be clear that in doing this we are adding some
regulatory burden. Fortunately, the procedures we have outlined are
noninvasive and non-risky, so it is likely that the amount of
oversight needed for the project would be relatively minimal.

Bruce