EMBL, Heidelberg and Max-Delbrück-Centre for Molecular Medicine
The large amount of data created by world-wide sequencing efforts calls for automation in data handling and analysis. This requires accurate storage and updating mechanisms as well as appropriate retrieval software. While molecular databases are the most valuable source of information for comparative analysis, they are, like the accessing software, also the product of history and far from perfect. Thus, at present, working with sequence databases requires knowledge about their powers and their pitfalls (1,2). I will demonstrate both aspects by presenting various examples related to the prediction of protein function and comparative genome analysis.
More than 80% of all known genes have at least one identifiable homologue in current databases; for the majority of them functional predictions are possible. An example of such predictions is given for the breast cancer gene BRCA1 (3). This shows that sequence analysis is making a transition from a service performed after experimental characterisation of biological data to a role in guiding further experiments.
With the availability of complete gene pools of model organisms, we are becoming able to analyse their functional composition as well as to study the evolution of proteins and pathways. A drift of protein function from metabolism to regulation and communication during evolution can be observed. More "modern" eukaryotic proteins, involved in communication and regulation, tend to have a modular architecture i.e. consist of many structurally and often functionally independent building blocks that allow rapid evolution (4).
1. Bork, P. and Bairoch, A. (1996) Trends Genet. 12,
October issue
2. Bork, P. (1996) Science 271, 431-432
3. Koonin, E.V., Altschul, S.F. and Bork, P. (1996) Nat.
Genet. 13, 266-268
4. Bork, P., Downing, K.A., kieffer, B. and Campbell,
I.D. (1996) Quart. Rev. Biophys. 29,
119-167