TDF Format

The ".tdf" file format is an indexed, compressed binary file format to support the display of numeric data. It was developed simultaneously with, and is similar in purpose, to the UCSC "bigWig" format. For must use cases we now recommend the "bigWig" format is it is widely used across many tools, while "tdf" is limited in use to IGV.

TDF files are created with igvtools. Input formats include wig and bedgraph as well as the IGV specific ".igv" and ".cn" formats. TDF files representing alignment coverage can also be created directly from .bam files using igvtools.

General layout

Header
Tiles
Datasets
Groups
Master Index

Header

Field	Description	Type	Value
magic	TDF magic number	int	TDF\0
version	format version number	int
indexPosition	file position of index section	int
indexSize	size in bytes of index section	int
headerSize	size in bytes of the remainder of the header section	int
nWindowFunctions	number of window functions	int
List of window functions
windowFunction	window function name	string	mean median min max percentile2 percentile10 percentile90 percentile98 stddev count density
End list of window functions
trackType		string
trackLine	UCSC style track line	string
nTracks	number of tracks	int
List of track names (n = nTracks)
trackName	name of track	string
*End list
genomeId	genome identifier (e.g. hg19)	string
flags	Flags	int

Tiles

This section contains tiles of data. A tile represents a region of the genome at a specific zoom (resolution) level. Each tile is referenced by a tile index entry of a dataset.

Field	Description	Type	Value
type	tile format	string	fixedStep variableStep bed bedWithName
Remainder according to type

type=fixedStep

Field	Description	Type	Value
nPositions	Number of genomic positions	int
start	genomic start position (zero based)	int
span	genomic span for each data point	float
List of data points. Track order first. (n= nTracks X nPositions)
datum	data value for track and position	float
End list

type=variableStep

Field	Description	Type	Value
tileStart	genomic position for start of tile	int
span	genomic span for each data point	float
nPositions	Number of genomic positions	int
List of data start positions
start	genomic start position (zero based)	int
End list
List of data points. Track order first. (n= nTracks X nPositions)
datum	data value for track and position	float
End list

type=bed

Field	Description	Type	Value
nPositions	Number of genomic positions	int
List of data start positions. (n= nPositions)
start	genomic start position (zero based)	int
End list
List of data end positions. (n= nPositions)
end	genomic end position	int
End list
nSamples	Number of samples. Ignored
List of data points. Track order first. (n= nTracks X nPositions)
datum	data value for track and position	float
End list
Optional feature names (type = bedWithName)
List of feature names (n=nPositions)
name	feature name	string
End list

Dataset

A dataset is a container for tiles of data at a given zoom level. Tiles are referenced by file position.

Field	Description	Type	Value
nAttributes	Number or attributes	int
List of attributes
key	Attribute key	string
value	Attribute value	string
End list
dataType	ignored	string
tileWidth	Width of each tile in base pairs	float
nTiles	Number of tiles	int
List of tile entriee
position	File position for start of tile	long
size	Size of tile in bytes	int
End list

Group

A Group is a container of key-value pairs, essentially a dictionary. A TDF file can in theory have an arbitrary number of groups referenced from the group index. In practice only a single group, the "root group" with name "/", has been used. The root group contains meta data and statistics for the file as a whole. See below for common attributes.

Field	Description	Type	Value
nAttributes	Number or attributes	int
List of attributes
key	Attribute key	string
value	Attribute value	string
End list

Master Index

Field	Description	Type	Value
nDatasets	Number of datasets	int
List of datasets
name	dataset name	string
position	dataset file position	long
nBytes	size of dataset in bytes	int
End list
nGroups	Number of groups	int
List of groups
name	name of group	string
position	file position of group	int
nBytes	size of group in bytes	int
End list

Root group attributes

TDF files created with igvtools typically include the following attributes in the root group (group name = "/"). The data type for all attributes is "string".

Name	Description
2nd Percentile	2nd percentile value of all data in this file
10th Percentile
90th Percentile
98th Percentile
Maximum
Mean
Median
Minimum
chromosomes	Comma delimited list of all chromosomes/contig/sequence names in this file
maxZoom	The maximum pre-computed zoom level
totalCount	For alignment coverage files only - total number of alignments.