While the best instructions on velvet are in the documentation, for basic use all you need is the following (Note: run this on psoda64.cs.byu.edu - it won't work on psoda4 or dna or the other machines):
bash$ velveth outputDirectory 21 fragments.fa
bash$ velvetg outputDirectory
This will create outputDirectory and put the assembled contigs into outputDirectory/contigs.fa.
The '21' has to do with the size of the hash (sec. 5.1 in the Velvet manual), and there are other options for optimizing the algorithm on short and long fragments with -short and -long. These two commands should get you along for this project, though please feel free to look at the documentation and play around with velvet's options if you want.
Working Example
The following example should get you going. Copy everything from ~clement/cs418/ch5 into your own directory so you can change things.
>original
actaggacattagagacata
gggtttatacaggattaaaa
>NODE_1_length_10_cov_2.700000
ACATAGGGTTTATA
>NODE_2_length_8_cov_4.750000
ACATTAGGACAT
>NODE_3_length_8_cov_3.500000
GATTACAGGATT
CLUSTAL 2.0.12 multiple sequence alignment
original ACTAGGACATTAGAGACATAGGGTTTATACAGGATTAAAA
NODE_2_length_8_cov_4.750000 ------ACATTAG-GACAT---------------------
NODE_1_length_10_cov_2.700000 ------ACATAGGGTTTATA--------------------
NODE_3_length_8_cov_3.500000 -------GATTACAGGATT---------------------
** *