Compare CDDM and PSPA in Tyr kinase

from katlas.core import *

import pandas as pd,seaborn as sns

Load CDDM

cddm = Data.get_cddm()

cddm = cddm.iloc[:,:-6]

Load PSPA

pspa = Data.get_pspa_tyr_norm().iloc[:,:-6]

Match PSPA index to CDDM

pspa_info = pd.read_csv('raw/lew_tyr_info.csv')

pspa

	-5P	-5G	-5A	-5C	-5S	-5T	-5V	-5I	-5L	-5M	...	5H	5K	5R	5Q	5N	5D	5E	5s	5t	5y
kinase
ABL1	0.0668	0.0689	0.0646	0.0520	0.0564	0.0539	0.0485	0.0448	0.0520	0.0536	...	0.0613	0.0652	0.0756	0.0526	0.0512	0.0362	0.0339	0.0254	0.0254	0.0337
TNK2	0.0679	0.0818	0.0627	0.0617	0.0529	0.0528	0.0419	0.0463	0.0437	0.0453	...	0.0499	0.0385	0.0302	0.0531	0.0465	0.0630	0.0572	0.0364	0.0364	0.0572
ALK	0.0675	0.0640	0.0590	0.0511	0.0476	0.0422	0.0455	0.0514	0.0546	0.0543	...	0.0448	0.0367	0.0489	0.0334	0.0387	0.0245	0.0226	0.0181	0.0181	0.0172
ABL2	0.0687	0.0715	0.0611	0.0448	0.0537	0.0513	0.0467	0.0398	0.0462	0.0505	...	0.0566	0.0640	0.0779	0.0538	0.0565	0.0378	0.0381	0.0252	0.0252	0.0289
AXL	0.0656	0.0753	0.0535	0.0525	0.0468	0.0467	0.0459	0.0538	0.0507	0.0542	...	0.0441	0.0506	0.0355	0.0635	0.0696	0.0592	0.0559	0.0413	0.0413	0.0455
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
KDR	0.0634	0.0672	0.0556	0.0517	0.0541	0.0526	0.0427	0.0420	0.0428	0.0476	...	0.0543	0.0653	0.0771	0.0509	0.0582	0.0414	0.0387	0.0335	0.0335	0.0406
FLT4	0.0457	0.0531	0.0488	0.0553	0.0512	0.0471	0.0432	0.0499	0.0474	0.0530	...	0.0624	0.0564	0.0559	0.0537	0.0610	0.0620	0.0528	0.0600	0.0600	0.0464
WEE1_TYR	0.0531	0.0640	0.0559	0.0560	0.0433	0.0435	0.0568	0.0571	0.0637	0.0562	...	0.0585	0.1058	0.1658	0.0447	0.0495	0.0312	0.0365	0.0453	0.0453	0.0490
YES1	0.0677	0.0571	0.0537	0.0530	0.0527	0.0505	0.0435	0.0375	0.0400	0.0463	...	0.0593	0.0662	0.0840	0.0559	0.0604	0.0422	0.0482	0.0374	0.0374	0.0411
ZAP70	0.0602	0.0880	0.0623	0.0496	0.0471	0.0514	0.0465	0.0380	0.0307	0.0526	...	0.0484	0.0477	0.0290	0.0520	0.0537	0.0709	0.0710	0.0862	0.0862	0.0605

93 rows × 230 columns

Get overlapped kinase

# non-overlapped
pspa.index[~pspa.index.isin(cddm.index)]

Index(['BMPR2_TYR', 'DDR1', 'ERBB2', 'LIMK1_TYR', 'LIMK2_TYR', 'MAP2K4_TYR',
       'MAP2K6_TYR', 'MAP2K7_TYR', 'PKMYT1_TYR', 'NEK10_TYR', 'PDHK1_TYR',
       'PDHK3_TYR', 'PDHK4_TYR', 'PINK1_TYR', 'TESK1_TYR', 'TNNI3K_TYR',
       'WEE1_TYR'],
      dtype='object', name='kinase')

overlap_kinase = pspa.index[pspa.index.isin(cddm.index)]
overlap_kinase

Index(['ABL1', 'TNK2', 'ALK', 'ABL2', 'AXL', 'BLK', 'PTK6', 'BTK', 'CSF1R',
       'CSK', 'MATK', 'DDR2', 'EGFR', 'EPHA1', 'EPHA2', 'EPHA3', 'EPHA4',
       'EPHA5', 'EPHA6', 'EPHA7', 'EPHA8', 'EPHB1', 'EPHB2', 'EPHB3', 'EPHB4',
       'BMX', 'PTK2', 'FER', 'FES', 'FGFR1', 'FGFR2', 'FGFR3', 'FGFR4', 'FGR',
       'FLT3', 'FRK', 'FYN', 'HCK', 'ERBB4', 'IGF1R', 'INSR', 'INSRR', 'ITK',
       'JAK1', 'JAK2', 'JAK3', 'KIT', 'LCK', 'LTK', 'LYN', 'MERTK', 'MET',
       'MST1R', 'MUSK', 'PDGFRA', 'PDGFRB', 'PTK2B', 'RET', 'ROS1', 'SRC',
       'SRMS', 'SYK', 'TEC', 'TEK', 'TNK1', 'NTRK1', 'NTRK2', 'NTRK3', 'TXK',
       'TYK2', 'TYRO3', 'FLT1', 'KDR', 'FLT4', 'YES1', 'ZAP70'],
      dtype='object', name='kinase')

overlap_column = pspa.columns

pspa = pspa.loc[overlap_kinase,overlap_column]

cddm = cddm.loc[overlap_kinase,overlap_column]

Pearson

from matplotlib import pyplot as plt

d = pspa.corrwith(cddm,axis=1).sort_values(ascending=False)

pspa.corrwith(cddm,axis=1).sort_values(ascending=False)

kinase
BLK       0.422068
TNK2      0.418332
TXK       0.366034
LCK       0.364885
ABL2      0.355762
            ...   
MATK      0.028732
FLT4      0.020456
PTK6     -0.001976
FES      -0.006169
PDGFRA   -0.040095
Length: 76, dtype: float64

sns.set(rc={"figure.dpi":200, 'savefig.dpi':200})
sns.set_context('notebook')
sns.set_style("ticks")

pspa.corrwith(cddm,axis=1).sort_values(ascending=False).plot.bar(figsize=(15,3))
plt.xlabel('')
plt.ylabel('Pearson');

Compare PSSM heatmaps

from matplotlib import pyplot as plt
from katlas.plot import *
import os, seaborn as sns
from tqdm import tqdm
from PIL import Image

def plot_kinase(df,kinase, title, fname=None, aa_order_paper = [i for i in 'PGACSTVILMFYWHKRQNDEsty']):
    
    # get PSSM matrix
    m = get_one_kinase(df, kinase, drop_s = False).T.loc[aa_order_paper]

    # plot heatmap
    plot_heatmap(m,title=title)
    
    
    plt.savefig(fname,bbox_inches='tight', pad_inches=0.05) if fname else plt.show()
    plt.close()

def plot_cor(k,title, fname=None):
    
    # plot
    plot_corr(cddm.loc[k],pspa.loc[k],'CDDM','PSPA')
    plt.title(title)
    
    plt.savefig(fname,bbox_inches='tight', pad_inches=0.2) if fname else plt.show()
    plt.close()

kinase_order = pspa.corrwith(cddm,axis=1).sort_values(ascending=False).index

kinase_order

Index(['BLK', 'TNK2', 'TXK', 'LCK', 'ABL2', 'TEC', 'ABL1', 'FYN', 'EPHA6',
       'SRMS', 'EPHB4', 'DDR2', 'EPHA5', 'EPHB2', 'YES1', 'EPHB3', 'HCK',
       'SYK', 'EPHA7', 'FGFR4', 'EPHA2', 'EPHA8', 'EPHA4', 'PTK2B', 'ITK',
       'LYN', 'FRK', 'PTK2', 'CSF1R', 'JAK3', 'JAK1', 'SRC', 'LTK', 'EPHB1',
       'AXL', 'MST1R', 'MERTK', 'EPHA1', 'FLT1', 'EGFR', 'FER', 'KIT', 'BMX',
       'INSRR', 'FGFR3', 'JAK2', 'EPHA3', 'MET', 'BTK', 'FLT3', 'FGR', 'KDR',
       'ERBB4', 'ZAP70', 'RET', 'TNK1', 'FGFR2', 'TYK2', 'FGFR1', 'ALK',
       'PDGFRB', 'TEK', 'MUSK', 'TYRO3', 'INSR', 'NTRK3', 'ROS1', 'NTRK1',
       'NTRK2', 'IGF1R', 'CSK', 'MATK', 'FLT4', 'PTK6', 'FES', 'PDGFRA'],
      dtype='object', name='kinase')

# get count of KS pairs in CDDM
df = Data.get_ks_dataset()

# Convert substrate names to uppercase
df['SUB'] = df['substrate'].str.upper()

# Remove duplicates based on kinase and substrate
df_unique = df.drop_duplicates(subset=['kinase_paper', 'SUB'])

# Count unique substrates for each kinase
cnt_unique = df_unique.groupby('kinase_paper').size()

# Count number of substrates (with duplicates) for each kinase
cnt_general = df.kinase_paper.value_counts()

Uncheck below to run all

for k in tqdm(kinase_order):
    
    # print('CDDM')
    plot_kinase(cddm,k,title= f'{k} from KS datasets (n={cnt_unique[k]})')
    
    # print('PSPA')
    plot_kinase(pspa,k, title = f'{k} from PSPA')
    
    plot_cor(k,title = k)
    
    break

  0%|          | 0/76 [00:00<?, ?it/s]

  0%|          | 0/76 [00:00<?, ?it/s]

To save, uncomment below

# for k in tqdm(kinase_order):
    
#     # print('CDDM')
#     plot_kinase(cddm,k,title= f'{k} from KS datasets (n={cnt_unique[k]})', fname=f'corr_tyr/CDDM/{k}.png')
    
#     # print('PSPA')
#     plot_kinase(pspa,k, title = f'{k} from PSPA', fname=f'corr_tyr/PSPA/{k}.png')
    
#     plot_cor(k,title = k, fname=f'corr_tyr/pear/{k}.png')
    
#     # break

Combine images

def combine_images_custom_layout(image_paths, output_path):
    images = [Image.open(image_path).convert('RGBA') for image_path in image_paths]
    
    # Calculate total width and height for the new image
    total_width = max(images[0].width, images[1].width + images[2].width)
    total_height = images[0].height + max(images[1].height, images[2].height)
    
    # Create a new image with calculated dimensions
    combined_image = Image.new('RGBA', (total_width, total_height))
    
    # Paste the first image at the top-center
    x_offset = (total_width - images[0].width) // 2
    combined_image.paste(images[0], (x_offset, 0), images[0])
    
    # Paste the second image at the bottom-left
    combined_image.paste(images[1], (0, images[0].height), images[1])
    
    # Paste the third image at the bottom-right
    combined_image.paste(images[2], (images[1].width, images[0].height), images[2])
    
    # Save the result
    combined_image.save(output_path)

Uncomment below to save combined figure for pdf

# folders = ["corr_tyr/pear",'corr_tyr/CDDM','corr_tyr/PSPA']

# for k in tqdm(kinase_order,total=len(kinase_order)):
#     filename = f"{k}.png"
#     image_paths = [os.path.join(folder, filename) for folder in folders]
    
#     output_path = f"corr_tyr/combine/{k}.png"
#     combine_images_custom_layout(image_paths, output_path)
    
#     # break

Convert images to pdf

Correlation of amino acids DE and phospho-S/T/Y

import pandas as pd

from katlas.core import *
from katlas.plot import *

import seaborn as sns
from matplotlib import pyplot as plt

cddm = Data.get_cddm()

pspa= Data.get_pspa_tyr_norm().iloc[:,:-6]

# remove dual
pspa = pspa[pspa.index.str.split('_').str.len() ==1]

overlap_kinase = cddm.index.intersection(pspa.index)

overlap_columns=pspa.columns

pspa = pspa.loc[overlap_kinase,overlap_columns]

cddm = cddm.loc[overlap_kinase,overlap_columns]

D = cddm.columns[cddm.columns.str.contains('D')].tolist()

E = cddm.columns[cddm.columns.str.contains('E')].tolist()

DE = D+E

cddm[DE].max(1)

s = cddm.columns[cddm.columns.str.contains('s')].tolist()
t = cddm.columns[cddm.columns.str.contains('t')].tolist()
y = cddm.columns[cddm.columns.str.contains('y')].tolist()

sty = s+t+y

sns.set(rc={"figure.dpi":200, 'savefig.dpi':200})
sns.set_context('notebook')
sns.set_style("ticks")

plt.figure(figsize=(5,4))
plot_corr(cddm[DE].median(1), pspa[sty].max(1),'CDDM DE median', 'PSPA sty max')

plt.figure(figsize=(5,4))
plot_corr(cddm[E].median(1), pspa[sty].max(1),'CDDM DE median', 'PSPA sty max')

plt.figure(figsize=(5,4))
plot_corr(cddm[E].median(1), pspa[y].max(1),'CDDM DE median', 'PSPA sty max')

plt.figure(figsize=(5,4))
plot_corr(cddm[E].median(1), pspa[s].max(1),'CDDM DE median', 'PSPA sty max')

plt.figure(figsize=(5,4))
plot_corr(cddm[DE].median(1), pspa[t].max(1),'CDDM DE median', 'PSPA s/t max')

plt.figure(figsize=(5,4))
plot_corr(cddm[E].median(1), pspa[y].max(1),'CDDM E median', 'PSPA y max')

plt.figure(figsize=(5,4))
plot_corr(cddm[D].median(1), pspa[sty].max(1),'CDDM DE median', 'PSPA sty max')

To save data, uncheck below

# data_list = [cddm[DE].median(1),
#              cddm[D].median(1),
#              cddm[E].median(1),
#              pspa[t].max(1),
#              pspa[y].max(1)]

# d = pd.concat(data_list,axis=1)

# d.columns=['CDDM_DE_median','CDDM_D_median','CDDM_E_median',
#            'PSPA_t_max','PSPA_y_max']

# d.to_csv('source/Fig4CD.csv')