todo.md

In cmvcordova/histoneSig:

Immediate

• Get a robust, working implementation of signal_feature_matrix
• Investigate why functions break when equal signals are provided (mapply), and when single observation signalsets are provided.
• Add parameter to signal_feature_matrix for different genomes based on BSgenome-

Can wait a bit

• Wrap signal_feature_matrix to genomic ranges
• PONDER ABOUT adding an additional slot to the signalset class, "from experiment". This would have the goal of being able to feed a signalsetlist to signal_feature_matrix and automatically create columns per experiment. OR a way to join only the experiment-exclusive signal column from signal_feature_matrix objects (Figure out which is more or less computationally expensive).
• np_signals_from_bigwig should have a way to read the queried intervals only, instead of loading all of the bigwig file into memory and then segment from it.

Long Term

• Add functionality to plotSignal to highlight regions that overlap regions of interest (ex: If plotting ChIP signal, show which ranges overlap with ATAC if predicting open chromatin).
• When positions instead of indices are specified in plotSignal, adjust x axis to show this.
• See if theres an object that can compactly store signals
• If above fails, HDF5 File compatibility for dampening the cost associated to representing signals in a long format
• Build a wrapper for being able to read multiple file pairs for cell/tissue type experimenting.

Thoughts that could probably culminate in some useful implementation:

• Can you build a general accesor function or redesign signalset classes to stop having to lapply every single parameter at the start of a function?

cmvcordova/histoneSig documentation built on Sept. 14, 2020, 12:02 a.m.