LeucipPy : A Protein Geometry Correlation Python Library

Currently it can be installed from Test PyPi

pip install -i https://test.pypi.org/simple/ LeucipPy-pkg-RachelAlcraft

The library relies on these Bioinformatics librararies:

• BioPython for protein structure coordinates. (Cock et al, 2009; Hamelryck et al, 2003)

Future versions will also use:
• PDB EDA for loading the electron density. (Yao et al, 2019)
• DSSP for asigning secondary structure. (Joosten et al, 2015; Kabsch & Sander, 1983)

There are also the common dependencies of matplotlib, seaborn, numpy and pandas

Not all these are implemented during transition to PyPi, and there are some more eg hexbins.

1.Scattter Plot	2.Probability Density Plot	3.Histogram
scatter	probability	histogram
		count mean std min 25% 50% 75% max 525.0 1.48 0.01 1.44 1.47 1.48 1.48 1.51
4.Pdb Data Report	5.Electron Density Report	6.Contact Map
DataView	DensityView (not currently implemented)	Contacts
7.Probability Density Difference Report
Difference

• Correlation reports
• Contact Maps
• 1000 Structures
• Bimodal Tau example for each amino acid
• Correlation reports show structure with suspicious geoemtry
• Above structure with reports that match ProCheck
• Looking at the electron density against geoemtry and coordinates part 1
• Looking at the electron density against geoemtry and coordinates part 2

from LeucipPy import LeucipPy as leu

• leu.getEbiLink(pdb_code): Returns the html path to the pdb file on the ebi server
• leu.getElectronDensityLink(pdb_code, diff=False): Returns the html path to the 2FoFc or FoFc electron density file on the ebi server
• leu.getPdbLink(pdb_code): Returns the html path to the pdb file on the ebi server
• leu.getStructureFactorLink(pdb_code): Returns the html path to the structure factors file on the ebi server
• leu.getEnghHuberStatistics(type='BackboneBondLengths'): Returns a dataframe of the E&H summary statistics from 1991

You need to first create a list of biopython objects. This downloads from the servers for you.
(Nb this is code to create biopython structures.)

import os
from urllib.request import urlretrieve
import Bio.PDB as bio
from LeucipPy import LeucipPy as leu
########
pdb_codes = ['1egj','3nir'] #A list of whatever pdb codes you wish to look at
########
parser = bio.PDBParser()
strucs = []
for pdb_code in pdb_codes:    
  pdb_file, pdb_html_loc = leu.getPdbLink(pdb_code)
  print(pdb_file, pdb_html_loc)  
  if not os.path.exists(pdb_file):
      urlretrieve(pdb_html_loc, pdb_file)
  struc = parser.get_structure(pdb_code,pdb_file)
  strucs.append(struc)
 

Use LeucipPy to generate a dataframes of geoemtric correlations

 from LeucipPy import GeoDataFrame as gdf
 
 ########
 geos = ['C:O','N:CA']
 hues = ['bfactor','aa']
 #######
 geo = gdf.GeoDataFrame(strucs)
 df = geo.calculateGeometry(geos,hues)

• gdf.GeoDataFrame(biopython_structures) : The class is initialised with a list of biopytong structures
• gdf.calculateGeometry(self,geos,hues=['pdb_code','resolution','chain','aa','rid','ridx','bfactor','occupancy']): Pass a list of geoemtric measures and a list of hues (which defaults to all of them as shown).
• Geos - distance: Any 2 atom types, colon delimitted, e.g. N:CA is the backbone bond distance between N and CA, CA:CA+1 is the c-alpha distance between this and the next CA. If you pass in CA:CA it will return the magnitude vector of the atom position. This is useful for comparing structures that had been adjusted.
• Geos - angle: Any 3 atoms will give an angle, eg N:CA:C gives the backbond angle TAU and CA-1:CA:CA+1 gives the CA angle.
• Geos - dihedral: Any 4 atoms gives a dihedral, standard like PHI and PSI as C-1:N:CA:C and N:CA:C:N+1 or anything at all, eg c-alpha as CA-2:CA-1:CA:CA+1

This class is a shortcut to formatting reports and they are saved in an html format. The transition to PyPi is underway so not all reports are currently available. Currently:

from LeucipPy import GeoHTML as ghm

rep = ghm.GeoHTML('LeucipPy Colab Report','ColabLeucippy.html')
rep.addPlot2d(df,'scatter','C:O','C:N+1','bfactor')
rep.addPlot2d(df_data,'seaborn','atom_no','bfactor','element') 
rep.addLineComment('1 dimensional data below')
rep.addPlot1d(df,'histogram','C:O')
rep.addDataFrame(df['C:O'].describe(),'C:O')
rep.addBoxComment('Produced by LeucipPy')
rep.printReport()
  

• ghm.GeoHTML(title,outpath, cols=3) : The class is initialised with a main title, the outpuf file path, and the number of columns in the html table
• ghm.addPlot2d(data,plottype,geo_x,geo_y,hue,title='',palette='viridis'): Pass it a dataframe, a plottype, the columns for x and y, and optionally a title and a colour palette.
  --plottype can be: scatter,seaborn
• ghm.addPlot1d(data,plottype, geo_x,title='',palette='crimson'):Pass it a dataframe, a plottype, the column for x, and optionally a title and a colour palette.
  --plottype can be: histogram
• ghm.addDataFrame(data,title=''): Format a dataframe in a box as a grid.
• ghm.addSeries(data,title='',transpose=False): Format a series in a box as a grid (aimed at describe() data).
• ghm.addLineComment(comment): Break the table to add a horizontal comment
• ghm.addBoxComment(comment): Add a comment in a table box
• ghm.changeColNumber(cols): Breaks the current html table and starts a new one with a different number of columns