The answer would be related to have tools to integrate the biological data in a common framework offering, at the same time, computational technologies (Java, SQL ....) to perform complex queries over the integrated biological data.

If you are interested on this approach have a look to JBioWH (http://jbiowh.googlecode.com) and send to us your "biological questions". We will try to find an answer for you.

Thank you for your interesting answers! So actually, do you think future tools should/could help the scientist to shape "good questions" ? Think about this for example: in finances the investors sometimes use "Financial market analysis" to help decide their future investments, could a scientist use bioinformatics tools for decision-making ?

Saying as a bio-data interpreter of RNA viral populations rs I think there is one concern that needs to be taken care before start to sequence everything.

What that large amount of data that is the output of that new sequencing generation mean to the organism in question ?.

A serious question in HCV new treatments is if a resistant variant that appear in a really low frequency can really cause viral breakthrough. When sequencing HCV from patients using new techniques the resistant variants can be seen in really low frequencies but the failure of the treatment and the fixation of the resistant variant in these patients appear to not be direct related to that low presence. If we use this techniques in the clinical practice we will deny treatment to a number of patients that in the future would not have developed resistance.

More than never the Understanding of the amount of data of new generation sequencing will depend on several factors, not only the interpretation of the Bioinformatic tools output.

Dear German, you are right that "the human capacity of interpretation of the information" is the most important step on any research project. However, in this moment the biologists has a lot of questions that we (bioinformaticians) are unable to respond due to many reasons. One of the first reason is that we depend so much of the big data centres (NCBI, EMBL, GenBank, KEGG, ........), its web tools and its limitations. The actual Bioinformatic point of view, which is the use of this big data centres for everything, reduce the possibilities to answer important questions using the system biology point of view (with all the biological data integrated).

The second reason is the huge explosion on available biological data. This is the present and next challenge on Bioinformatic. We need develop software to handled that explosion using an integrative point of view. The tools will have to handled a huge amount of data and, at the same time, perform complex queries.

Nowadays, the personal computers are powerful. You can easily have 1TB of hard disk, 16 GB of RAM and a powerful 8 core processor in a laptop. So, the researchers could execute heavy computational procedures on his PC. However, the Bioinformatics tools were designed with the traditional point of view and we are wasting a lot of computational power.

So, my point is that we need to develop new bioinformatics tools with a new perspective and offering to the users (researchers) the possibility to work in any moment. We need to be prepared for the complex questions that we face everyday in our labs.

You mentioned Ingeniuty and Metacore. Those two are so good not because their unique data structures or data integration/refining techniques. They are so good - because they are heavily and manually curated by human experts who are paid for doint the curation work. This is what made them successful startups. GeneGo must have sold their Metacore business to Thomson Reuters for a decent price...

That type of tools are somewhat natrually entering now the area of medicine. The key point may be to make the medical doctors understand and trust them. The psychological element that "there is a human reasoning" behind each element of knowledge-based system is even more important for the medics than for the molecular biologists.

When accuracy of biological experiments will be evidenced, tools will be usefull. For now, they are just enough to "predict" what you expect to see... Biological experiments, then tools, then good combination... In the meantime, informatic toold can just heklp to make mistake in interpretations....

you guys are bringing up very important points, there is obviously so much to improve in bioinformatics so it could one day produce significant contributions. Sometimes, like some of you pointed out, bioinformatics don't make more than confusing us. In the other hand, it is hard for me to imagine a future in which computer won't be an important part of biological science and medicine. So the question is, how can do we get there?

I think quite often the analysis of biological experiments becomes an exercise in story-telling; one hopes to see moves towards making some predictions that can be measures and verified. I would love to see research papers making more concrete and measurable results statements that could be databased more directly, and tools aggregating and making use of these predictions. Ingenuity and Metacore of course already distill such results, but it is done by people reading the papers and making judgements about how to in fact interpret and store the results.

I would also love to see the reproducible science initiative catch on more in biology/bioinformatics, with researchers submitting the code that generates the results along with their data, and tools evolving that can store both the data and the code in some joint ways, making it easier for people to test and easily recover the results.

I fully agreed with Dana, that there must be coded with the data available to scientist with which they can first they can validate it and then do there own analysis. As reproduciblity is a major concern with majority of tool not achieving this target. Bioinformatics tools should be developed along these lines.

Thank you guys!, how many of you tried Ingenuity and MetaCore? what is it that made you like/dislike them? . Just in case, I don't work for any company, just trying to get ideas in general ;)

I'd like to mention a free alternative to Ingenuity IPA and Metacore, it's called ConsensusPathDB and integrates interactions and human-curated pathways from overall 30 open-access databases. It can help you analyze a list of genes or metabolites in several ways based on pathways, GO categories, interactions, etc. Give it a try: http://ConsensusPathDB.org .

I think it will be key for these future tools to incorporate a rich semantic layer, so data can be interpreted within a specific context. This would of course mean that data repositories would also be annotated (with rich ontologies, capable of supporting reasoning) and related data would be linked across repositories. I believe a shared semantic structure would support the integration of smaller data banks with the big ones most tools use today.

I miss a tool that represents any experiment in a structured way. I believe for the sake of reproducible science, the materials and methods in any paper should be coded in a database-entry, related with the PMID of that paper. The coding keys of course should be widely-accepted.

It's just frightening to see phrases like "the liver was homogenized in lysis buffer and the proteins were separated by SDS-PAGE and then transferred to nitrocellulose" and I would say such kind of phrases are more probable to be seen in the journals with high impact factor. I believe that the method used in some determination biases the final result, but for now I am not aware of any tool that can quantify that bias and predict the probability of false-positive or false-negative result.

It is amazing how far the omics technologies already have reached, and how difficult it could be to just understand how someone extracted the DNA, verified the cells viability or verified that the results obtained by a physiological measurement in an animal model are not due to the stress (provoked by the treatment), but by the treatment itself.

Another important database would be the "NS tool".

What we see in the papers are mostly the statistically significant differences between one treatment and another. Below the tip of that iceberg of "significant" results lies a huge mountain of experiments that show no differences. That is a very important knowledge that stays silent in the notebooks and never reaches the journals.....I realize though that this idea would require a quite different approach when subimiting a paper. What a fortune of money, animals, manhours and enthusiasm will be saved if people share their "no differences" results in a structured way!

Thank you guys for the interesting comments. Petar, as for your second comment there is this thing http://www.jnrbm.com/ is for real!