VALIDATING AN ILLUMINA SAMPLE SHEET

import pandas as pd

csv_in = '2020-05-14_NovaSeq.csv'

CREATE A DataFrame CALLED df FROM THE CSV

# Create a pandas DataFrame from CSV 
# Formatting all data as strings since all data is qualitative
df = pd.read_csv(csv_in, dtype='str')

# Get rid of the empty "NA" rows that Excel loves to include...
df.dropna(inplace=True)

# Print some summary statistics using "f-strings"
print(f'Rows:   {len(df)}')
print(f'Plates: {df.plate.nunique()}\n')

# Preview the first 5 rows
print(df[:5])

Rows:   384
Plates: 4

  sample-number plate well rawSample-name         sample-name
0             1     1  A01    15-4-SW-2-1         15-4-SW-2-1
1             2     1  A02    15-4-SW-3-2         15-4-SW-3-2
2             3     1  A03    17-100-SW-A  17-100-SW-A-1-1-37
3             4     1  A04    17-100-SW-A  17-100-SW-A-1-3-30
4             5     1  A05    17-100-SW-B  17-100-SW-B-1-1-30

CHECKING SAMPLE COUNT PER PLATE

• Each plate should have 96 rows

plates = df['plate'].value_counts()
bad_plates = plates[plates != 96]
try:
    assert len(bad_plates) == 0
    print(f'[PASS] All {len(plates)} plates have 96 samples!')
except AssertionError:
    print(f'[FAIL] Found {len(bad_plates)} of {len(plates)}', 
          f'that do not have 96 samples...')
    print(bad_plates)

[PASS] All 4 plates have 96 samples!

CHECKING FOR DUPLICATE SAMPLE NAMES

• sample-name should be unique

duplicates = df[df['sample-name'].duplicated()]
try:
    assert len(duplicates) == 0
    print('[PASS] All sample names are unique!')
except AssertionError:
    print(f'[FAIL] Found {len(duplicates)} duplicate sample names...')
    print(duplicates)
    

[PASS] All sample names are unique!

COMPARING THE SAMPLE NAME WITH THE RAW SAMPLE NAME

• sample-name should start with rawSample-name

bad_rows = df[[row['sample-name'].startswith(row['rawSample-name']) == False 
               for i, row in df.iterrows()]]                                            
try:
    assert len(bad_rows) == 0
    print('[PASS] Every sample-name starts with its rawSample-name!')
except AssertionError:
    print(f'[FAIL] Found {len(bad_rows)} rows', 
          'where sample-name does not contain rawSample-name')

[PASS] Every sample-name starts with its rawSample-name!

CHECKING THE LENGTH OF SAMPLE NAMES

• sample-name should be 100 characters or less

bad_rows = df[df['sample-name'].str.len() > 100]
try:
    assert len(bad_rows) == 0
    print('[PASS] All sample names are 100 characters or less!')
except AssertionError:
    print(f'[FAIL] Found {len(bad_rows)} bad rows...\n')
    print(bad_rows)

[PASS] All sample names are 100 characters or less!

CHECKING THE FORMATTING OF SAMPLE NAMES

• sample-name should only contain: [a-z], [A-Z], [0-9] or '-'
• sample-name should not start or end with '-'

pattern = r'^[a-zA-Z0-9]+[a-zA-Z0-9\-]+[a-zA-Z0-9]$'
bad_rows = df[df['sample-name'].str.match(pattern) == False]
try:
    assert len(bad_rows) == 0
    print('[PASS] All sample names are properly formatted!')
except AssertionError:
    print(f'[FAIL] Found {len(bad_rows)} bad rows...')
    print(bad_rows)

[PASS] All sample names are properly formatted!

Everything appears to be in order...