How to normalize microarray data using Python?

Question

[melissachua90]

I have an Affymetrix microarray dataset to analyse time-dependent treatment on gene expression in different patients.

I have three types of controls:

• positive control
• negative control
• probeset control

My attempt:

import qnorm
import pandas as pd


# Negative normalization: Subtract the median of the negative controls within the patient sample
neg_ctl_median = neg_ctl_df.iloc[:,-29:].median()

# Positive normalization: Divide the median of the positive controls within the patient sample
pos_ctl_median = pos_ctl_df.iloc[:,-29:].median()

# Probeset normalization (quantile normalization)
probeset_norm = qnorm.quantile_normalize(pos_ctl_df.iloc[:,-29:], axis=1)

for gene, sample in coding_gene_df.iloc[:,-29:].astype(float).iterrows():
  norm_val = sample - neg_ctl_median # Negative normalization
  norm_val = norm_val / pos_ctl_median # Positive normalization (replace sample value with the value that has already been normalized against negative control)
  norm_val = norm_val / probeset_norm

Is my code on the right track - especially the probeset normalization? Should I divide the samples by probeset mean?

The data (as dictionary)

treatment_df.head().to_dict()

{'0h_P1_T1_TimeC1_PIDC4_Non-Survivor': {'DNAJB6 /// TMEM135': '9.25',
  'DNAJC14': '8.44',
  'DNAJC15': '8.66',
  'DNAJC30': '7.34',
  'DNAJC9': '7.62'},
 '0h_P2_T1_TimeC2_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '9.26',
  'DNAJC14': '8.34',
  'DNAJC15': '8.63',
  'DNAJC30': '7.42',
  'DNAJC9': '7.19'},
 '0h_P3_T1_TimeC1_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '9.07',
  'DNAJC14': '8.67',
  'DNAJC15': '8.91',
  'DNAJC30': '7.15',
  'DNAJC9': '7.22'},
 '0h_P4_T1_TimeC1_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '8.56',
  'DNAJC14': '8.63',
  'DNAJC15': '9.29',
  'DNAJC30': '7.46',
  'DNAJC9': '7.16'},
 '0h_P5_T1_TimeC4_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '9.42',
  'DNAJC14': '8.32',
  'DNAJC15': '8.63',
  'DNAJC30': '7.29',
  'DNAJC9': '7.29'},
 '12h_P1_T4_TimeC2_PIDC4_Non-Survivor': {'DNAJB6 /// TMEM135': '8.58',
  'DNAJC14': '8.64',
  'DNAJC15': '8.98',
  'DNAJC30': '7.2',
  'DNAJC9': '7.23'},
 '12h_P2_T4_TimeC3_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '8.94',
  'DNAJC14': '8.38',
  'DNAJC15': '8.03',
  'DNAJC30': '7.32',
  'DNAJC9': '7.26'},
 '12h_P3_T4_TimeC2_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '9.27',
  'DNAJC14': '8.57',
  'DNAJC15': '8.27',
  'DNAJC30': '7.32',
  'DNAJC9': '7.39'},
 '12h_P4_T4_TimeC2_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '8.5',
  'DNAJC14': '8.25',
  'DNAJC15': '8.47',
  'DNAJC30': '6.98',
  'DNAJC9': '7.12'},
 '12h_P5_T4_TimeC2_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '9.18',
  'DNAJC14': '8.25',
  'DNAJC15': '8.66',
  'DNAJC30': '7.38',
  'DNAJC9': '7.6'},
 '24h_P1_T5_TimeC4_PIDC4_Non-Survivor': {'DNAJB6 /// TMEM135': '8.13',
  'DNAJC14': '8.43',
  'DNAJC15': '8.33',
  'DNAJC30': '7.41',
  'DNAJC9': '6.68'},
 '24h_P2_T5_TimeC3_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '8.85',
  'DNAJC14': '8.54',
  'DNAJC15': '8.73',
  'DNAJC30': '7.26',
  'DNAJC9': '7.49'},
 '24h_P3_T5_TimeC3_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '8.99',
  'DNAJC14': '8.42',
  'DNAJC15': '8.92',
  'DNAJC30': '7.25',
  'DNAJC9': '7.27'},
 '24h_P4_T5_TimeC3_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '9.26',
  'DNAJC14': '8.6',
  'DNAJC15': '7.62',
  'DNAJC30': '6.93',
  'DNAJC9': '7.3'},
 '24h_P5_T5_TimeC3_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '9.31',
  'DNAJC14': '8.14',
  'DNAJC15': '8.11',
  'DNAJC30': '7.52',
  'DNAJC9': '7.64'},
 '48h_P1_T6_TimeC3_PIDC1_Non-Survivor': {'DNAJB6 /// TMEM135': '9.05',
  'DNAJC14': '8.27',
  'DNAJC15': '7.63',
  'DNAJC30': '7.46',
  'DNAJC9': '7.42'},
 '48h_P2_T6_TimeC3_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '8.57',
  'DNAJC14': '8.23',
  'DNAJC15': '8.6',
  'DNAJC30': '7.45',
  'DNAJC9': '7.62'},
 '48h_P3_T6_TimeC3_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '9.23',
  'DNAJC14': '8.65',
  'DNAJC15': '7.71',
  'DNAJC30': '6.77',
  'DNAJC9': '7.26'},
 '48h_P4_T6_TimeC3_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '8.99',
  'DNAJC14': '8.7',
  'DNAJC15': '7.99',
  'DNAJC30': '7.28',
  'DNAJC9': '7.5'},
 '48h_P5_T6_TimeC3_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '8.78',
  'DNAJC14': '8.21',
  'DNAJC15': '8.02',
  'DNAJC30': '7.56',
  'DNAJC9': '7.69'},
 '4h_P1_T2_TimeC1_PIDC4_Non-Survivor': {'DNAJB6 /// TMEM135': '8.25',
  'DNAJC14': '8.69',
  'DNAJC15': '8.78',
  'DNAJC30': '7.53',
  'DNAJC9': '7.29'},
 '4h_P2_T2_TimeC2_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '9.49',
  'DNAJC14': '8.47',
  'DNAJC15': '8.7',
  'DNAJC30': '7.41',
  'DNAJC9': '7.23'},
 '4h_P3_T2_TimeC2_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '9.87',
  'DNAJC14': '8.38',
  'DNAJC15': '8.03',
  'DNAJC30': '7.29',
  'DNAJC9': '7.57'},
 '4h_P5_T2_TimeC2_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '9.2',
  'DNAJC14': '8.1',
  'DNAJC15': '7.9',
  'DNAJC30': '7.47',
  'DNAJC9': '7.62'},
 '8h_P1_T3_TimeC4_PIDC4_Non-Survivor': {'DNAJB6 /// TMEM135': '8.49',
  'DNAJC14': '8.58',
  'DNAJC15': '8.22',
  'DNAJC30': '7.29',
  'DNAJC9': '7.13'},
 '8h_P2_T3_TimeC2_PIDC2_Survivor': {'DNAJB6 /// TMEM135': '9.16',
  'DNAJC14': '8.13',
  'DNAJC15': '7.88',
  'DNAJC30': '7.58',
  'DNAJC9': '7.18'},
 '8h_P3_T3_TimeC4_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '8.52',
  'DNAJC14': '8.44',
  'DNAJC15': '9.23',
  'DNAJC30': '7.38',
  'DNAJC9': '7.2'},
 '8h_P4_T3_TimeC4_PIDC1_Survivor': {'DNAJB6 /// TMEM135': '9.28',
  'DNAJC14': '8.53',
  'DNAJC15': '8.22',
  'DNAJC30': '7.32',
  'DNAJC9': '6.84'},
 '8h_P5_T3_TimeC4_PIDC3_Survivor': {'DNAJB6 /// TMEM135': '9.29',
  'DNAJC14': '8.41',
  'DNAJC15': '9.08',
  'DNAJC30': '7.28',
  'DNAJC9': '7.69'}}

[ATpoint]

Affymetrix Microarray Analysis In Python

[melissachua90]

The link you shared did not have any info about normalization