All Linux and OS distros have python 3.x, no need to convolute the code with python 2.x backwards compatibility

Follow https://docs.readthedocs.io/en/latest/conda.html to swith from virtualenv to conda when building the documentation in readthedocs

Release instructions in the wiki

lmfit does not assume any weights when calculating residual, we need to pass the inverse of the experimental errors as weights

Some property class like SecondaryStructureProperty and SaxsProperty can be instatiated using an
executable command through a method call. For instance:

saxs_prop = SaxsProperty().from_crysol_pdb(pdb_file)

will create X-ray property saxs_prop by running executable crysol using PDB file pdb_file.
Analogously,

ss_prop = SecondaryStructureProperty().from_dssp_pdb(pdb_file)

will create secondary structure property ss_prop by running executable dssp using PDB file pdb_file.

The task is to create a utility function so that given: (1)a property class, (2) a method, and (3) a list of files will create one property for each file, and this should be done in parallel.

For instance,

import multiprocessing
from tqdm import tqdm
with multiprocessing.Pool(processes=multiprocessing.cpu_count()) as pool:
    profiles = list(tqdm(pool.imap(SaxsProperty().from_crysol_pdb, pdb_names),
                         total=len(pdb_names)))

will create list of saxs properties profiles using all available cores.

The utility function should be implemented in file utils.py

We need to do the same as in the the qef repository with a requirements.txt file and build in a virtual environment.

Should be a profile property. Every residue should have a real number indicating the expected number of bonds to form.
Utsab used VMD for this.

Description

k-means clustering on a trajectory chunk.

Show representative structure and secondary structure for fitted nodes

Create a "feature vector" using some of the properties of a node. To be used later to create an array from which to calculate distances between feature vectors with scipy.spatial.distance_matrix or sklearn.metrics.pairwise.euclidean_distances

Description

Reach for the node attribute in a ScalarProperty and plot a histogram of the set of y-values obtained from the leafs under the node.

Do a fit at a particular cluster depth

For the example notebooks, insert a check for installation of crysol. If a Not-Found error is raised, then fetch the data from the helper idpflex_data repository.

• should use dssp
• for every residue, save a dictionary like {'T': 0.12, 'H': 0.34} indicating the probability of the residue to
  be in a turn (T), helix (H), and so on. Probabilities should be normalized.

• idpflex version:
• Python version:
• Operating System:

Description

Describe what you were trying to get done.
Tell us what happened, what went wrong, and what you expected to happen.

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.

Use all available cores

When creating artificial experimental X-ray profile in the example notebooks, we select a particular level of the tree and then assign weights to the nodes. For instance, for file hiAPP_clustering.ipynb we have in cell 13:

weights = [0.00, 0.13, 0.0, 0.55, 0.0, 0.32, 0.00] # the weights add up to one

Then in after executing cell 18 we obtain

('n192_amplitude', '0.32')
('n189_amplitude', '0.00')
('n187_amplitude', '0.00')
('n181_amplitude', '0.00')
('n191_amplitude', '0.00')
('n188_amplitude', '0.13')
('n190_amplitude', '0.55')

Containing three nodes with non-zero weigths, as expected.

If instead we use

weights = [0.00, 0.13, 0.0, 0.55, 0.0, 0.00, 0.32] # the weights add up to one

The we obtain

('n192_amplitude', '0.00')
('n189_amplitude', '0.00')
('n187_amplitude', '0.00')
('n181_amplitude', '0.00')
('n191_amplitude', '0.00')
('n188_amplitude', '0.13')
('n190_amplitude', '0.55')

There's no node with weight 32 !
This happens allways when we assign a non-zero weight to the last item of list weights.

RMSD calculation does not perform the rotation to minimize this quantity

• idpflex version:
• Python version:
• Operating System:

Description

Describe what you were trying to get done.
Tell us what happened, what went wrong, and what you expected to happen.

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.

This is a matrix property, not a profile one.
Each entry will be a real number indicating the probability that two residues are in contact.
There should be additional information indicating how the contact is defined. If implemented with MDAnalysis, optional arguments could be provided when creating the property.

For the example notebooks, insert a check for installation of dssp. If a Not-Found error is raised, then fetch the data from the helper idpflex_data repository.

"The PCA+K-means clustering was done on R. But earlier, I had done only K-means clustering in python. Please find the attached python script. The packages are on the script" - Utsab

K-means-clustering-3D.py.zip, rg-eed-sasa.txt.zip

Depends on #66

1. Obtain centroids by RMSD of each trajectory chunk
2. Cluster the centroids by RMSD or by custom distance matrix

Three kinds of plotting:

1. node
2. leafs
3. fractions

node plotting will create a 2D colored plot where each column denotes a particular element of secondary structure and each row denotes a particular residue. The entries on each row should be normalized to one, and colored from blue (zero) to one (red).

leafs will plot the secondary structure assignments for each leaf of the node. Leafs will be sorted according to distance to the secondary structure assignment of the node.

Fractions plotting will create a bars plot, with each bar representing the fraction of certain secondary structure in the protein. Fractions will be created from the secondary structure assignment of the node.

Tentative To-Do's:

• folder with recipes
• wiki documentation

• idpflex version:
• Python version:
• Operating System:

Description

Describe what you were trying to get done.
Tell us what happened, what went wrong, and what you expected to happen.

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.

Describe the need and motivation behind writing this software

For changes that don't deserve an issue + pull-request

Description

Derive from ScalarProperty, calculate from PDB using MDAnalysis

Use the script I did for Utsab's trajectories.

Also, fix the multiprocessing loading of crysol files in parallel when working in python 2.7

Function to create a cluster of certain number of leafs, each with a random distance from zero to one. Then cluster according to this random distance

When you calculate the e-e-d using gmx polystat, you will find the eigenvalues (say L1, L2 and L3) of radius of gyration tensor in the last three columns of the output file. Then, we can just use the standard asphericity formula as used in Loukas's JACS paper (eq 1). I wrote very simple shell script to calculate that for each frame. I think it will be much easier for you to implement in python. Here's the link for that paper

Bucket issue for new releases

Should produce a matplotlib plot

After introduction of clustering and fitting with lmfit

• idpflex version:
• Python version:
• Operating System:

Description

Describe what you were trying to get done.
Tell us what happened, what went wrong, and what you expected to happen.

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.