In the dataset we collected, the sequenced eukaryotic genomes ranged from 373 to 3,175,581 thousand base pairs (kbp) in size, while the genomes of non-eukaryotes (including bacteria, archaea, viruses, mitochondria, and chloroplasts) were substantially smaller, i.e., 2.4 9949.9 kbp (or kilobases in the case of single-stranded viral DNA or RNA) (Figure 1A).
To generate the grape dataset we collected GBS data from a collection of diverse samples from the genus Vitis including commercial Vitis vinifera varieties, hybrids and wild accessions from the USDA grape germplasm collection.
In the second dataset, we collected as diverse datasets of similar and dissimilar interfaces as we could reliably get using a structure-based similarity measure.
Evaluation dataset: for the evaluation dataset, we collected 6 months of PubMed log data (February, April, May, June, July and August, 2009), consisting of more than 333 million user queries, 329 million abstract views in 144 million user sessions.
Training dataset: for the training dataset, we collected 2 months of PubMed log data (March 2008 and February 2009), consisting of more than 100 million user queries and 130 million abstract clicks in 51 million user sessions.
For future work, we will calibrate the results obtained from this study, where reaction time was analyzed, with the wider massive dataset we collected during the experiment.
For the flu dataset, we collect tweets containing the keyword 'flu'.
The main difference between the two datasets we collected lies in the different encoding of video traces.
From each of the 49 datasets we collected genes up-regulated in stem cells into a stem cell gene list (SGL) and, where available, a corresponding set of genes up-regulated in differentiated cells into a differentiated gene list (DGL) (Table S1 and Dataset S2).
To test how these generalize to the large datasets we collected we have reformulated some of these observed trends as possible explanations to the low conservation rates and analyzed them using our integrated networks.
Moreover, to verify the effectiveness of the adaptive threshold with large datasets, we collected streams totally containing 90 changes for evaluation.
