Personal tools
You are here: Home / Consensus

README

Consensus calculation with aria2.3

Concept:
1) perform 20 independent ARIA runs with identical data and parameters except random seed.
2) generate a set of consensus restraints by merging final assignments for cross-peaks kept until the last iteration in <n>% of the runs

First download the aria2.3.2 archive (and install it) and the aria_consensus.py script


How to:

0) Prepare an ARIA project with the parameters you would use for a single ARIA run (aria_project.xml)

1) use the aria_consensus.py script to generate the 20 new ARIA projects (will automatically change random seed and run numbering)
python aria_consensus.py --generate --n 20 aria_project.xml

will generate aria_project_c[0-19].xml

2) Set up the 20 calculations, e.g. in bash command line:
for p in aria_project_c{0..19}.xml ; do aria2 -s $p; done

3) run the 20 calculations (might take some time...)

4) generate the consensus restraints using a threshold of 60% to consider peaks as usable for the consensus
python aria_consensus.py --merge --n 20 --cutoff 0.6 aria_project.xml

will generate:
- consensus_0.6.xml (the restraints)
- aria_project_consensus.xml (the ARIA project)

Now run the aria_project_consensus.xml calculation.

---------------------

Usage (requires the ARIA2 environment variable to be set)

aria_consensus.py --generate|merge [--n 20] [--cutoff 0.6] aria_project.xml

    --generate         Generate ARIA XML projects (using aria_project.xml) as
                       reference with different random seeds numbered from 1 to <n>

    --merge            Generate consensus restraints (XML format) from <n> runs

    --n <integer>      Number of runs to generate/merge (default: 20)

    --cutoff <float>   Cut-off for keeping consensus cross-peaks (default: 0.6)


References:

1) Buchner L, Güntert P. Increased reliability of nuclear magnetic resonance protein structures by consensus structure bundles. Structure. 2015 Feb 3;23(2):425-34. doi: 10.1016/j.str.2014.11.014. Epub 2015 Jan 8. PMID: 25579816.

2) Alexander Belyy, Florian Lindemann, Daniel Roderer, Johanna Funk, Benjamin Bardiaux, Jonas Protze, Peter Bieling, Hartmut Oschkinat, Stefan Raunser. Mechanism of threonine ADP-ribosylation of F-actin by a Tc toxin. b

« September 2022 »
September
MonTueWedThuFriSatSun
1234
567891011
12131415161718
19202122232425
2627282930