Network servers are inexpensive ways to provide services to researchers at all institutions regardless of size or available funding. Many laboratories have only PCs or Macs; when larger departmental-level facilities are available, they typically are charged for CPU and access time. In this proposal we will implement a multiple sequence alignment network server providing protein and nucleic acid multiple sequence alignment. The user will have a choice of Pattern Induced Multiple Alignment (PIMA) for protein sequences or CLUSTAL V for nucleic acid or protein sequences. Initially, access will be via simple e-mail; later, a Unix-based client-server socket-level program will be provided, then a Gopher+ access mechanism. In this way, researchers who have only e-mail will be able to access the multiple sequence alignment server. Those with full Internet connections will be able to access the server more conveniently. Client-side source code will be provided to developers so that users of the Genetic Data Environment, SeqApp, GCG, and other software packages will be able to use this server as """"""""pull-down"""""""" or other integrated part of their analysis environment. We also will continue to characterize d2, a fast sequence similarity search algorithm by D. Torney, C. Burks, D. Davison, and K. Sirotkin. It has the unique advantage of word weighting as an integral part of the algorithm. We will investigate methods of determining optimal sensitivity and selectivity. We will also simplify the procedure for determining optimal word weighting and extend our current results on estimating the variance of a d2 for word weighting. Lastly, the hiring of a programmer on the multiple sequence alignment server will also help to continue operation of the University of Houston Gene-Server, which has been providing molecular biological data and software via e-mail, Gopher, and WAIS since 1990.
