| |
| | 9. The buildmodel estimation process |
 | | Sequence weighting is particularly important when, as normal, the sequence data given to SAM is biased toward some type or subfamily of sequences (for example, from those organisms that have been most studied). |
 | | Sequence weighting is particularly important in database discrimination: without sequence weighting, the model may specialize to an over-represented subset of the sequences, meaning that family members that do not happen to be in that sub-family will receive low scores. |
 | | By default, internal weighting of alignments is turned on, with Henikoff relative weighting, 0.5 target bits per column, and a 0.5 exponent. |
| www.sacs.ucsf.edu /Documentation/sam/node11.html (4260 words) |
|