python_cheatsheet.md

Python Cheatsheet

Pandas

NumPy islcose()

• Incredibly useful for things like matching mz, rt info between mzmine, metatlas, etc.

mz = 123.4567
# find all detected m/z within .001 daltons
df[np.isclose(df['mz'], mz, atol=.001)]
# could alternately calculate atol with my ppm functions for better matching

String methods reminders

• Useful options are case=False and regex=False. Turning regex off is sometimes critical to avoid errors, but then you can't search for multiple patterns using contains().

• .str.contains()

  • uses '|' operator in between patterns as an OR statement, only works if regex=True
  • I do not know of a way to do an AND statement within the function

• .str.startswith() and .str.endswith()

  • Both of them use tuples to separate patterns with an OR statement

Decorators

From https://pabloariasal.github.io/python-decorators-from-the-ground-up/

Example decorator for timing a function

import time

# this is the "decorating" function
def timer(func):
    def inner_func(*args, **kwargs):
        start = time.time() # set up timing
        func_out = func(*args, **kwargs) # do the funciton
        print time.time() - start # print the log
        return func_out # return the function's output
    return inner_func # return the function's output

# decorate a function
@timer
def test_func(a, b, c='jedi'):
    return a + b + c

# use the function as normal
>>> test_func('the', 'last')
1.19209289551e-06
'thelastjedi'

Argument Parsing

import argparse
# parse arguments
parser = argparse.ArgumentParser()
# a couple examples
parser.add_argument('-l', '--level', help='how many levels deep to search the chemical network', type=int, choices=[1,2,3], default=2, required=True)
parser.add_argument('--mute', help='mutes pandas warnings', action='store_true')
# store arguments
args = parser.parse_args()
# use arguments
args.level
args.mute

Multiprocessing

This opens a number of python processes, which on some systems can cause memory issues because it appears that the amount of memory used has increased by the number of processes open. I'm sure there's a way to work around this.

import multiprocessing as mp
max_cpu = mp.cpu_count()
p = mp.Pool(max_cpu)
out = p.map(func, iterable)
p.close()
p.terminate()

Matplotlib

First, some of my favorite matplotlib colors to use when you only need to use a few colors:

coral, yellowgreen, lightskyblue, gold, gray, black

Basic plotting

import matplotlib.pyplot as plt
plt.rcParams['pdf.fonttype'] = 42 # makes editing text in Illustrator much easier
%matplotlib inline # only when in an Ipython notebook
# it is always better to use axes instead of plt for more control of 
# multiple axes and/or multiple plots
ax1, fig = plt.subplots() 
ax1.plot(x, y, color=, label=, marker=, linestyle=, linewidth=)
ax1.set_ylabel('label')
ax1.set_xlabel('label')
plt.title('title')
plt.show()

color bar

From https://stackoverflow.com/questions/8342549/matplotlib-add-colorbar-to-a-sequence-of-line-plots

fig, ax = plt.subplots()
... # whatever plot you want to make
sm = plt.cm.ScalarMappable(
	cmap='viridis', # any color map
	norm=plt.Normalize(vmin=0, vmax=1) # vmin and vmax are your map's min and max
	)
sm._A = []
cbar = fig.colorbar(sm, orientation='vertical')
cbar.ax.set_ylabel('tanimoto score')
cba.ax.set_yticks() # can do normal axes functions this way too

Fun axes functions

ax1.set_yscale('log') # try to capture any y-values that are zero, otherwise they won't be plotted
ax1.set_ylim(low, high)
ax2 = ax1.twinx() # for separate y-axes
ax2.plot(x, y)

Tick labels

# using plot object
plt.xticks(tick_positions, labels)

# using axes object
ax.set_xticks(tick_positions)
ax.set_xticklabels(labels)

# rotation (kwargs are the same for plt or ax)
# set 'ha' to 'right' to align right when rotating between 0 and 90
plt.xticks(tick_positions, labels, rotation=45, ha='right')

Basic legend stuff

• This will place the legend box outside the plot, with the upper left corner of the plot just to the right of the upper right corner of the plot

ax1.legend(loc='upper left', bbox_to_anchor=(1.1, 1.))

The loc argument says how to orient the bbox_to_anchor, and bbox_to_anchor is in units of x and y axes positions: (0,0) is bottom left, (1,1) it top right

• This will make a legend most useful for bar charts

import matplotlib.patches as mpatches
handle_list = []
sample_colors = ['red', 'green', 'blue']
for i, label in enumerate(['control', 'sample_1', 'sample_2']):
    patch = mpatches.Patch(color=sample_colors[i], label=label)
    handle_list.append(patch)
ax.legend(handles=handle_list, loc='lower left', bbox_to_anchor=(-1., 0))

Setting figure size

fig.set_size_inches(12,6) # (width, height)

Subplots

It is really useful to ravel subplots, so you can act on them individually

fig, axs = plt.subplots(rows, cols, sharex=True, sharey=True)
axs = axs.ravel()
for i, coordinates in list_of_coordinates:
	x = coordinates[0]
	y = coordinates[1]
	axs[i].plot(x, y)
	axs[i].set_title('individual subplot title')

### Adjust the spacing of subplots
fig.subplots_adjust(wspace=, **kwargs) # I find this to be way better than tight_layout(), and usually use wspace, but there are other kwargs that give you more control
fig.suptitle('title of overall plot', fontsize=24)

Fun error options

Scatterplot with error bars

ax.errorbar(x, y, yerr=, xerr=, fmt=data_marker, color=fmt_color, ecolor=error_bar_color, *kwargs) # many cool kwargs to customize error bars

Showing error via shading (good for time series)

ax.plot(x, y) # main data
ax.fill_between(x, y-error, y+error, color=, alpha=0.2)