rnajena / vegeta Goto Github PK

write conda recipe and build docker container

Clustering doesn't make much sense when there are only 10 to 20 sequences present in the original input file. Output a warning message stating that the alignment step is done w/o initial clustering.

Use dinucleotide shuffling on individual structural elements to test for significance (z-score analysis)

When huge amounts of data (tested: PB2 of IAV, >60.000 sequences) are given as input,
pickle complains about the size.
Chunking the dataset seems to be the best idea here.

Apparently there are some problems with TABs when using JalView and/or RNAplot.

Apparently, sometimes the file containing representative sequences is completely empty, causing VeGETA to exit.

Making an alignment of the overall input doesn't work in many cases as the sequences and the underlying structures are too diverse to be summarized in ONE alignment based on some sequences.
On top of this alignment, the individual clusters can be re-clustered -> from there, alignments of sub-clades are possible, which give a higher resolution.

This may lead to 20 different alignments for huge input sets though.

Implement a GoI (for example, an unpublished lab strain) which has to end up in the alignments.

ILP calculation takes quite some time; there are some local structures in the alignments that are independent of all other structures -- thus they can be solved individually. The ILP can be broken down into smaller ILPs, which reduces the overall complexity and thus run-time of the pipeline.

Implement a dry-run parameter that invokes all dependencies to check whether everything is correctly installed.

• loop size smaller than 4
• lonely pairs