I am guessing you are talking about RECOMB, one of the most important computational biology conferences. The story behind this is a little complicated and can show you some of the difficulties faced by a relatively new interdisciplinary field.
Since computational biology has both computer scientists and biologists, it needs to answer the needs of both cultures. For computer scientists, presentation and publication in the proceedings of an important conference is considered excellent. In biology, however, publication is mainly in journals. A second difference between cultures is that in computer science it is standard to publish extended versions of conference papers in journals. In biology, this is usually not acceptable unless it is a low-tier journal.
So what do you do if you want to get submissions of great work from computational biologists of both sides? This is a challenge that computational biologists have been dealing with not just in this conference but also in other conferences. There have been all kinds of proposed solutions - I will not list them here.
RECOMB organizers have tried different solutions and decided to call the accepted manuscripts "extended abstracts", and these are published only in the conference proceedings (which are not freely available). This way, computer scientists consider it published in the conference and biologists consider it unpublished so they can still submit it to a high-tier journal.
Actually they usually also have other "submission tracks" such as coordinated submission to a specific journal in parallel with the conference.