fitevd <query length> [options]
fitevd fits an extreme value distribution to a set of score-length pairs.
The outer loop alternates between estimating lambda and K with H fixed and estimating H with lambda and K fixed. The inner loop is the Newton-Raphson algorithm, which is used for solving the estimation problems.
Reads white-space separated score-length pairs from standard input.
The query length; ignored for -exp and -gcexp.
Writes the EVD parameters to standard output.
Option | Parameter | Description | Default Behavior |
---|---|---|---|
General Options | |||
-exp | Compute an exponential-Guassian mixture. | Compute EVD. | |
-gcexp | Compute a GC-dependent exponential-Guassian mixture. | Compute EVD. | |
-s | s | Size of length groups. | The size of length groups is 10,000. |
-ms | ms | Minimum number of scores per length (or GC-content) group. Note: If ms is 0 and -gcexp is given, then no binning is done and weighted regression is used. | The size of length groups is 200. |
-lr | lr | Minimum ratio of shortest to longest in group. | The minimum ratio of shortest to longest in the group is 1.50. |
-r | n | Use n randomly selected scores. | All are used. |
-p | Print p-values. | The p-values are not printed. | |
-h | H | Initial value of H. | H is set to 1.0. |
-ilh | ilh | Maximum iterations in outer L-H loop. | The maximum iterations in outer L-H loop is 10. |
-inr | inr | Maximum iterations in inner N-R loop. | The maximum iterations in inner N-R loop is 20. |
-d | smin smax | Print length vs p-value for smin ≤ score ≤ smax. | Do not print length vs. p-value. |
-t | Turn on debug tracing. | Debug tracing is not on. |