Microarray track

This document attempts to explain the microarray tracks on the browser in pretty extensive detail. The gene sorter has a somewhat different strategy for microarrays.

Database Representation

Microarray tracks are are stored as BED 15 data. A typical gapped BED item will have 12 fields. The microarray, or "expRatio" BEDs are another three fields tacked on to the gapped BED. These fields are the expCount, expIds, and expScores fields.

The following is an example BED 15:

#chrom	chromStart	chromEnd	name	score	strand	thickStart	thickEnd	reserved	blockCount	blockSizes	chromStarts	expCount	expIds	expScores
chr1	159639972	159640031	2440848	500	-	159639972	159640031	0	1	59,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	0.593000,1.196000,-0.190000,-1.088000,0.093000,-0.731000,0.130000,-0.008000,-1.087000,0.609000,-1.061000,-1.092000,0.807000,0.499000,-0.322000,-0.985000,0.309000,0.000000,0.812000,-0.457000,-0.560000,0.096000,0.186000,-1.092000,0.045000,0.573000,1.170000,1.336000,1.251000,1.919000,-0.056000,-0.189000,0.028000,
chr1	159640161	159640190	2440849	500	-	159640161	159640190	0	1	29,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	-0.906000,-1.247000,0.111000,-0.515000,-0.057000,-0.892000,0.167000,1.278000,0.051000,-0.596000,-0.251000,-0.826000,0.487000,0.714000,0.674000,1.046000,0.694000,0.236000,-0.718000,-1.196000,-1.274000,-1.278000,-1.055000,0.838000,-0.494000,1.137000,0.000000,0.690000,0.166000,-0.232000,0.174000,-1.253000,1.363000,
chr1	159640215	159640242	2440850	500	-	159640215	159640242	0	1	27,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	-0.465000,0.127000,1.215000,-0.073000,-0.465000,-0.141000,0.507000,-0.462000,-0.464000,0.570000,1.356000,0.559000,-0.459000,-0.464000,-0.458000,0.000000,0.322000,-0.454000,0.887000,-0.464000,1.196000,-0.463000,0.376000,-0.461000,0.547000,0.032000,-0.464000,0.066000,0.762000,-0.465000,-0.456000,0.919000,-0.464000,
chr1	159640256	159640309	2440851	500	-	159640256	159640309	0	1	53,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	1.779000,1.039000,-0.068000,-0.066000,-0.050000,0.665000,0.861000,-0.067000,0.000000,0.090000,-0.067000,1.240000,-0.018000,-0.068000,0.122000,0.478000,-0.068000,-0.068000,0.630000,-0.068000,0.092000,0.620000,-0.066000,-0.068000,1.601000,0.537000,1.103000,0.720000,1.959000,1.703000,-0.061000,-0.067000,1.097000,
chr1	159641139	159641250	2440852	300	-	159641139	159641250	0	1	111,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	0.000000,1.569000,-0.090000,0.330000,-0.091000,-0.089000,0.634000,1.319000,-0.090000,-0.092000,-0.048000,-0.091000,-0.091000,0.815000,1.455000,1.127000,1.549000,0.337000,0.432000,-0.092000,-0.092000,-0.090000,0.397000,1.082000,-0.092000,-0.091000,1.780000,2.025000,1.988000,2.575000,0.224000,-0.092000,-0.092000,
chr1	159642074	159642102	2440853	500	-	159642074	159642102	0	1	28,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	2.204000,-1.053000,-0.756000,1.135000,0.393000,0.366000,-0.730000,-0.113000,-1.179000,1.079000,-1.194000,-1.195000,-0.675000,1.124000,0.382000,-1.196000,0.313000,-0.380000,0.244000,-0.258000,2.070000,1.312000,-1.195000,1.901000,-1.191000,0.438000,-1.195000,1.096000,1.635000,0.232000,-0.440000,0.528000,0.000000,
chr1	159642586	159642611	2363661	600	+	159642586	159642611	0	1	25,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	0.486000,0.065000,0.897000,-0.566000,-1.169000,-1.547000,-0.266000,0.370000,-0.864000,-0.802000,1.214000,-0.240000,0.269000,-0.067000,0.375000,0.258000,0.366000,-0.275000,0.965000,0.785000,0.571000,0.511000,0.759000,0.658000,-0.196000,-0.055000,-0.017000,-0.411000,-0.450000,-1.028000,0.104000,-0.228000,-0.000000,
chr1	159642702	159642762	2363662	600	+	159642702	159642762	0	1	60,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	-0.772000,-0.532000,0.061000,-1.913000,-2.283000,-1.696000,-0.062000,0.200000,-0.609000,-2.054000,-1.184000,-2.465000,0.746000,0.673000,0.316000,0.041000,1.147000,0.590000,0.963000,0.919000,0.914000,0.000000,0.224000,-0.077000,0.395000,1.124000,0.690000,-0.369000,-1.427000,-0.970000,0.297000,0.335000,1.471000,
chr1	159642857	159642883	2363663	200	+	159642857	159642883	0	1	26,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	0.434000,-1.679000,-1.102000,0.529000,0.435000,0.614000,0.000000,0.175000,-1.517000,-1.148000,-0.180000,-1.379000,-0.798000,-0.163000,-0.219000,-1.936000,0.183000,0.627000,-0.933000,0.446000,1.150000,-2.162000,-0.018000,0.191000,1.351000,1.378000,-0.941000,0.743000,-1.632000,-0.415000,0.800000,-0.165000,1.257000,
chr1	159642959	159642988	2363664	600	+	159642959	159642988	0	1	29,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	0.000000,0.018000,-0.061000,-1.540000,-1.477000,-1.977000,-0.058000,-0.430000,-0.293000,-1.256000,-1.060000,-1.701000,0.715000,0.978000,0.980000,1.190000,1.248000,0.974000,0.907000,1.145000,0.782000,-0.152000,0.218000,-0.109000,1.550000,1.126000,1.537000,0.023000,-0.140000,-0.728000,0.520000,0.605000,0.597000,
chr1	159643028	159643054	2363665	600	+	159643028	159643054	0	1	26,	0,	33	0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,	0.291000,-0.093000,0.000000,-1.483000,-1.041000,-1.644000,0.219000,-0.189000,-0.466000,-0.942000,-0.832000,-1.026000,0.565000,0.868000,0.813000,0.786000,0.793000,0.574000,0.773000,0.764000,0.568000,-0.291000,-0.055000,-0.103000,0.916000,0.970000,0.993000,-0.106000,-0.230000,-0.402000,0.310000,0.433000,0.563000,

The expCount field says how many microarrays the track is using. The expIds field is used to link microarray labels to the microarray measurements in the expScores field. The microarray labeling information is not found in the BED 15 at all, rather it's part of the microarray configuration. Also, the expCount and expIds fields are constant across all rows, which may seem like a waste of space, but in theory the BED 15 spec allows more flexibility than it is actually given by its implementation in the browser. An example following the BED and trackDb is the following:

trackDb Settings

Microarray tracks in the browser must have several specific trackDb settings. The following is an example trackDb entry of a microarray track:

track affyHumanExon
shortLabel Affy All Exon
longLabel Affymetrix All Exon Chips
group regulation
priority 79.1
visibility hide
type expRatio
expScale 3.0
expStep 0.5
groupings affyHumanExonGroups

The important settings are: type, expScale, expStep, and groupings. "type expRatio" is what indicates to the browser that the track's data is a microarray. The expScale is a reflection of the dynamic range of the data, and is an absolute value. This effects the coloring of the track. An expScale of 3.0 will mean that most of the data is between -3.0 and 3.0 and that the brightest green will be used for values less than or equal to -3.0 and the brightest red for values greater than or equal to 3.0. The expStep value controls the color key step values on the deatils page. groupings sets the specific set of configurations to load from the microarrayGroups.ra file(s).

microarrayGroups.ra

The microarrayGroups.ra files are located in kent/src/hg/makeDb/hgCgiData in a directory/file structure similar to trackDb and hgNearData. These get copied to the apache cgi-bin where various CGIs read them directly. Like other configuration .ra files, these are meant to be relatively small compared to databases and be flexibly structured.

name affyHumanExonGroups
type groupings
all affyHumanExonAll
combine affyHumanExonGroupByTissueMean,affyHumanExonGroupByTissueMedian,
subset affyHumanExonSubsetByTissue,affyHumanExonSubsetByReplicate,
combine.default affyHumanExonGroupByTissueMedian 

name affyHumanExonAll
type all
description All Arrays
expIds 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,
groupSizes 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
names breast_A,breast_B,breast_C,cerebellum_A,cerebellum_B,cerebellum_C,heart_A,heart_B,heart_C,kidney_A,kidney_B,kidney_C,liver_A,liver_B,liver_C,muscle_A,muscle_B,muscle_C,pancreas_A,pancreas_B,pancreas_C,prostate_A,prostate_B,prostate_C,spleen_A,spleen_B,spleen_C,testes_A,testes_B,testes_C,thyroid_A,thyroid_B,thyroid_C, 

name affyHumanExonGroupByTissueMean
type combine mean
description Arrays Grouped By Tissue Mean
expIds 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,
groupSizes 3,3,3,3,3,3,3,3,3,3,3,
names breast,cerebellum,heart,kidney,liver,muscle,pancreas,prostate,spleen,testes,thyroid,

name affyHumanExonGroupByTissueMedian
type combine median
description Arrays Grouped By Tissue Median
expIds 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,
groupSizes 3,3,3,3,3,3,3,3,3,3,3,
names breast,cerebellum,heart,kidney,liver,muscle,pancreas,prostate,spleen,testes,thyroid,

Each paragraph of the microarrayGroups.ra file has the same basic fields name and type. In addition, with the exception of paragraphs with "type groupings", the paragraphs also have the description, expIds, groupSizes, and names fields.

The "type groupings" paragraph is linked to the trackDb setting "groupings" through the "name" setting. This paragraph will also define which other other paragraphs (perhaps subparagraphs) are connected with it. The all setting is required and points to the paragraph defining entirity of the arrays, ungrouped. The combine setting lists the paragraphs defining the other grouping methods for the set of arrays, and the combine.default mentions which of those is the default grouping method.

The "type all" and "type combine" paragraphs have similar formats. The description setting is required and it appears on the label of the track on the browser. The expIds correspond to the same expIds as the BED, but unlike the BED, these are meant to be in any order. The groupSizes and names settings both have the same number of words in the comma-separated lists and the order is important. In the affyHumanExonGroupByTissueMedian paragraph, the first of the names is "breast". The first of the groupSizes is "3", so the "breast" expIds are 0,1,2. Similarly, for cerebellum, the second groupSizes is 3, so those expIds are 3,4,5

Finally, if the paragraph type is "combine", then it requires and additional setting indicating how the arrays are combined. The two valid values are "median" and "mean".

Microarray Custom Tracks

The microarray custom track format is similar to a normal BED custom track format with the addition of some parameters to the "track" header which replace the trackDb and microarrayGroups.ra. Here's an example using the earlier example of a BED 15, but now as a custom track with 5 of the 33 arrays:

track type="array" expScale=3.0 expStep=0.5 expNames="breast_A,breast_B,breast_C,cerebellum_A,cerebellum_B," name="Microarray" description="Microarray custom track"
chr1	159639972	159640031	2440848	500	-	159639972	159640031	0	1	59,	0,	5	0,1,2,3,4,	0.593000,1.196000,-0.190000,-1.088000,0.093000,
chr1	159640161	159640190	2440849	500	-	159640161	159640190	0	1	29,	0,	5	0,1,2,3,4,	-0.906000,-1.247000,0.111000,-0.515000,-0.057000,

The expNames, expScale, and expStep settings are all required if the type="array".