Data Files

TSG suggested file format for experiment data

The TSG suggests a common file format for storing experimental data. Adhering to this format whenever practical makes it easier to re-use files and tools. The file is plain text for easy inspection and manipulation. The file format is a tab-separated values (tsv) file with the following specifications:

File

• File encoding is ASCII or UTF-8.
• The file contains no byte order mark (BOM) or other magic number. This makes it ASCII compatible.

Lines

• Lines are separated by the \r\n line delimiter for better compatibility between operating systems.
• The line delimiter should also be added after the last line. This simplifies stream reading since all records (lines) are terminated. This allows for the use of a readline() function for acquiring a line.
• The first line contains a header with column/field names.

Fields

• Field are separated by the tab field delimiter, because they rarely occur in texts and therefore require no escaping.
• The field delimiter should not be added after each line's last field. This allows for the use of a split() function for parsing a line.
• The last field in a line must not be empty, because it will show to parsers that the previous rule was obeyed.
• Fields are never surrounded by a quoting character.
• White space before or after field delimiters are considered part of a field.
• There is no defined escape character. If your data can contain tabs, use a different field delimiter or file format.

Data

• For numbers the decimal separator is a dot, not a comma. There is no thousands separator.

Example

An example of what a file in this format may look like:

User ID	Hair color	Response time	
1	brown	1.4	
2	blond	1230.434	
3	brown	0.399	

An example file can be downloaded here File:Example.zip (sorry, it is zipped).

Parsing

Importing such files can be done in many languages:

Python Standard Library

import csv
with open('example.tsv', 'rb') as csvfile:
    reader = csv.reader(csvfile, delimiter='\t', quoting=csv.QUOTE_NONE)
    for row in reader:
        print(', '.join(row))

or with header extraction

import csv
with open('example.tsv', 'rb') as csvfile:
    reader = csv.DictReader(csvfile, delimiter='\t', quoting=csv.QUOTE_NONE)
    print(', '.join(reader.fieldnames)) # print header
    for row in reader:
        print(', '.join([row[key] for key in reader.fieldnames]))

Note that when using Python 2 the field content will remain UTF-8 encoded (type=str). In Python3 strings will unicode (type=string).

Python Pandas

Pandas can interpret column type. You will have to store it separately or hardcode it.

import pandas as pd

d = pd.read_csv('example.tsv', delimiter='\t', skip_blank_lines=False, quoting=csv.QUOTE_NONE)

GNU R

d <- read.csv("example.tsv", head=TRUE, sep = "\t")

Alternatives

Data can be saved in a lot of file formats. If there is no reason to do otherwise, we prefer delimited files with the options shown in bold. Alternative options are also shown.

file extension	tsv	csv	dat	txt
file extension	ascii	UTF-8	UTF-16BE	UTF-16LE	UCS-4/UTF-32
magic number	None	<BOM>
line delimiter	\n	\r	\r\n
line delimiter after last line	no	yes
field delimiter	<tab>	,	;
field delimiter after last field	no	yes
quoting character	None	"	'
escape qc by doubling	no	yes
escape character	none	\
first line	contains header	contains data
last field in line	must not be empty	may be empty
whitespace following delimiter	part of field	not part of field
decimal separator	.	,
thousands separator	none	.	␣	U+2009

Note that tab characters and newlines cannot be present in field content.