The state of the program can be saved as an XML file, which contains all information of TFBS-map, gene and TF expression, as-well as the optimization settings and results.
== STREAM can now scan ==
You can provide the binding specificity of the TFs and scan a sequence with them directly, rather than using an external program to generate the TFBS-map. Stream supports two
formats a) count matrix (Jaspar format) and b) meme format. The sequence must be in
fasta format.
== Direction of the TFBSs are displayed ==
The direction (strand, anti-strand) of the TFBS can be displayed if the information was given.
== Joint training and visualization of multiple problems ==
A model can now be trained on more than one dataset. A dataset or "problem" has a TFBS-map, TF expression and Target expression information. You can combine any number of problems, as long as they are regulated by a similar set of TFs (no need to be exactly the same). The list of problems that are trained together and their properties are displayed in a table.
== Complexes can be generated ==
TF can co-operate and activate (or repress) transcription differently (e.g. stronger) if they have
formed a complex. The new version of STREAM therefore offers the user to specify which TFs can form a complex and what the task of the complex is. The concentration of the complex is
then the minimum of the expression levels of the two involved TFs. The expression levels of the individual TFs is then reduced accordingly. The binding sites of the complexes can be chosen to be
a) all sites of either TF (which means the TF-complex will compete with the individual TF for the TFBS)
b) new sites are generated for the complex at each location where the TFBS of both TFs are present within a certain distance
c) new sites at position where TFBSs exist of both TFs and no other TFBS in-between.
== Training can be stopped rather than canceled ==
Stopping the training will give you the current state of the optimization and the progress since start and stop is not lost.
Posts
