Add new metrics that Gabe suggested

[Haddox]

Counts of all possible rotamers
Dipeptide counts in sequence
Counts of all amino acid type pair interactions (serine residue or serine side chain within 4A of arg sidechain for example)
rosetta score of 2, 3, 4, 5 residue sequences - what is good/bad? normalized to what?
rosetta scores of small 3d neighborhoods
TERM-based analysis
counts of amino acids in certain secondary structures, or dipeptides in certain secondary structures
more granular fragment analysis, such as the quality of specific regions of the chain

[Haddox]

In commit 0373aa2, added code that computes amino-acid frequencies over all protein, in different secondary structures, and in different layers. Also computes di-peptide frequencies

[Haddox]

Counts of all possible rotamers
--Brian says no explicit function to do this. Could get list and then decide which one it's most like based on torsional angles

Counts of all amino acid type pair interactions (serine residue or serine side chain within 4A of arg sidechain for example)
--Alanine scan. Watch for other residues that change energy.
--Neighborhood residue selector then manual counting

rosetta score of 2, 3, 4, 5 residue sequences - what is good/bad? normalized to what?
--pose.energies().residue_total_energy(3) for site 3

rosetta scores of small 3d neighborhoods
--very hard to do it at the atom level, but could compute energies for all neighbors

TERM-based analysis

counts of amino acids in certain secondary structures, or dipeptides in certain secondary structures

more granular fragment analysis, such as the quality of specific regions of the chain
--Brian did this. The most important regions are the edges of loops (e.g., where a helix turns into a loop). Maybe could have different categories. Brian did this for 3mers using the following dssp pattern
--Bcov: dssp: xxHHHHHHHHhLLLLLhHHHHHHhLLLeEEEEEeLLLLxx

[Haddox]

In commit 6f8610b, added code for analyzing fragment quality at the level of individual sites (what is the average quality of all 9mer fragments centered upon a given site?) and within different kinds of secondary structures (what is the quality of all fragments that are centered upon a site within a given type of secondary structure?). Will add description of new metrics to README.

@grocklin: Added a few new metrics described briefly above. It looks like fragment quality is computed using 9mers. Did you ever have code to analyze the fragment quality of 3mers?

[grocklin]

I never did analysis of 3mers.

…

On Thu, Feb 14, 2019 at 12:33 PM Hugh Haddox ***@***.***> wrote: In commit 6f8610b <6f8610b>, added code for analyzing fragment quality at the level of individual sites (what is the average quality of all 9mer fragments centered upon a given site?) and within different kinds of secondary structures (what is the quality of all fragments that are centered upon a site within a given type of secondary structure?). Will add description of new metrics to README. @grocklin <https://github.com/grocklin>: Added a few new metrics described briefly above. It looks like fragment quality is computed using 9mers. Did you ever have code to analyze the fragment quality of 3mers? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AErvp7UOpuSmzijjQUfPZBDPZ-Y-NNXYks5vNcgbgaJpZM4aWepW> .

[Haddox]

Add average length of helices / strands / loops?
Fraction of internally satisfied bb hbond donors/acceptors?

[Haddox]

In commit 79f9b02, added code for analyzing per-residue Rosetta energies of fragments in primary sequence.

[Haddox]

In commit 72275b2, added code to compute energies of 3D neighborhoods around each residue

[Haddox]

In commit 8e2daa8, added code to compute the number of times each of the 20 amino acids contact each other in the structure