“Data sharing in the Biosciences II”
June 26 at 4:27 am
- Link
Going to start a new one as the previous one was getting long. - Cameron Neylon
Example of genome wide RNAi screens. Large scale datasets or large size. Could be seen as a pre-MIAME stage in comparison with DNA microarrays. Datasets can be hard to find, harder to analyse, serious hamburger issues. And nearly impossible to compare results from different labs - Cameron Neylon
RNAi globail initiative formed to bring researchers together. nearly thirty laboratories. Inforamtics Workgroup aims to develop tools to share and compare and analyse data. Challenges are to define which data to be shared, there is no current means to share it, therefore need standards and formats - Cameron Neylon
Important to distinguish between standards that specificy how to do an experiment and standards to describe how an experiment was done. Data standards are formal descirptions of the content of data sets. Metadata and raw data are both critical - not just processed data. - Cameron Neylon
Discussion of the importance of raw datasets. Suggestion that scientists want hamburgers not cows. Importance of re-checking the pdb with new refinement tools periodically mentioned so case for raw data robustly made. - Cameron Neylon
Data standards can also promote better practice in execution of experiments through SOPs and standardize the reporting of experiments. Processes need to be reasonably mature for this. Consistent use of procedures is required to make SOPs work. Segue into GLP and whether it applies to academic research - Cameron Neylon
Current practice is in the peer reviewed literature is insufficient to describe the experiment - Cameron Neylon
Did 1000 screens in ten labs with as close procedures as possible. Within group there is good technical reproducibility. Reproducibility between groups was poor. What is it we need to share to understand and define the differences? Built this based on standardisation exercise to accelerate development of standards. - Cameron Neylon
MIARE hosted at sourceforge. Issues of support for informatics in various groups. Development includes MIARE-TAB to enable excel based data input :) Need to harmonise standards -> MIBBI - Cameron Neylon