Massachusetts Institute of Technology researchers have developed a technique to derive fully comprehensive samples, or "sketches," of vast cellular datasets, to simplify analysis.
The team tweaked an algorithm designed to lay shapes over datasets, encompassing the full computational space with a "plaid covering" resembling a grid of equally-sized squares arrayed in multiple dimensions.
The algorithm only laid these multidimensional squares in places featuring one cell at minimum, bypassing empty spaces.
The final grid configuration had empty columns that were much wider or slimmer than the occupied columns, relieving computational load so the covering scaled to huge datasets.
The technique accelerated generation of sketches from massive datasets from hours to minutes, with the sketches distributing far more even representation of rare cells from across the datasets.
From MIT News
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA