WIXLIB

visualizing quantitative informationmartin krzywinski

outlinebest practices of graphical data designdata-to-ink ratiocartjunkcircosthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

graphical displays essentialsshow the datainduce viewer to think about substance rather than methodologyencourage eye to compare different pieces of dataavoid distorting what the data representspresent many numbers in a small spacemake large data sets coherentreveal data at several levels of detail – broad overview and fine structurethe visual display of quantitative informationedward r tufte, 2001, 2nd ed

graphics reveal data and patternseach of these sets are described by the same linear modelanscombe’s quarteteach of the values below is thesame for each setnumber of pointsaverage xaverage yregression linestandard error of slopesum of squaresqresidual sum of squarescorrelation coefficientr2the visual display of quantitative informationedward r tufte, 2001, 2nd ed

graphics organize complex informationsome data sets are naturally better represented visuallyeach of these data maps portrays 21,000 numbersalthough very dense,dense the images draw attention to hot spotsdeath rate from various cancersfemalesmalesthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

graphics organize dense informationlocations and boundaries,communes inof 330,000France240,000 numbersthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

graphics organize dense information1,024 x 2,222 sky divisions10 grey tonespixel grey value denotesnumber of galaxies incorresponding sky regiondensity of datacommensurate with aphotograph, butquantitativethe visual display of quantitative informationedward r tufte, 2001, 2nd ed

graphics simplify complex informationTGVthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

when the image is the datathe visual medium is idealf depictingford i ti multivariatelti i tdataarguably univariate andbivariate data should betabularized, withinreasonthis example shows a plotfor a case where datacannot be easilyparametrizedthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

parametrization of multivariate datathe 2D plane can depicthi h dihigh-dimensioni datad tchernoff faces are dataencodings designed foreasy identification ofoutliersparameters are mappedd tohead shape, eye distance,nose and lip sizesmoothly varying datacorresponds to smoothlyvarying chernoffpopulationlthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

data-to-inkdatato ink ratioproportion of graphic’s ink devoted to the non-redundant display of datai finformationti1.0 – proportion of a graphic that can be erased without loss of datainformationdata-to-ink ratio should always be maximized, within reasonthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

data-to-inkdatato ink ratiohighthe visual display of quantitative informationedward r tufte, 2001, 2nd edshockingly low

data-to-inkdatato ink ratiooriginaldeleted componentsmodified to increasedata-to-ink ratiothe visual display of quantitative informationedward r tufte, 2001, 2nd ed

shrink your graphicsdense data can be depicted within asmallll area withoutith t lossloff clarityl itas long as data-to-ink ratio is highgood graphics areinformativedensemultivariatestrive to give your viewerthe greatest number of ideasin the shortest timewithith the least inkin the smallest spacethe visual display of quantitative informationedward r tufte, 2001, 2nd ed

cartjunkexcessive use of grids and patterns cause perceived vibrationsavoid hatched patterns to limit moireavoid excessive use of decorative formsthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

the shimmering statisticnatural eye tremorandd densedfillpatterns produce ashimmering effectthis is annoying andtiringthe visual display of quantitative informationedward r tufte, 2001, 2nd ed

circosthere are manyy genomegbrowsers andvisualizers already available – do we reallyneed another one?communicating data visually critical forlarge data setsyp of data that obfuscatethere certain typescommon diagram formatsstandard 2D plots (2 perpendicular axes) are inadequate

scalar mappingsscalar valued mappings are common and easily handlediinputt genomici positioniti iis a scalarl iinputtwhen the output is real-valued (GC content, conservation, etc) use a histogram, line plot, scatter plotgenome position on x-axisfunction value on y-axisf :g y

genome-to-genomegenometo genome mappingsoutput scalar is often a genome position (G2G)range may beb theth same genome, or a differentdifft genomeG2G is also common, but less easily handledf : g g′genomepositiongenomeposition

drawing G2G mappings

drawing G2G mappingsGenome Res. 2003 Jan;13(1):37-45

drawing G2G mappingsGenome Res. 2003 Jan;13(1):37-45

drawing G2G mappingsGenome Res. 2005 May;15(5):629-40

drawing G2G mappingssc7sc15IIIchr04sIchr09ch

drawing G2G mappingsGenome Res. 2003 Jan;13(1):37-45

drawing G2G omics

drawing G2G /chr7paper/chr7data/030113/segmental/index.php

drawing G2G mappings

dealing with G2G mappingsreduce information content in figuresplot/colourmapl t/ ltargettt chromosome,hnott positionitif : g g ′ c′

dealing with G2G mappingsGenome Res. 2004 Apr;14(4):685-92

reduce samplingGenome Res. 2005 Jan;15(1):98-110

rearrange axes

partition data

recompose axis layout – circos

circoswritten in PerlApache-styleconfiguration fileplain text data inputpPNG output

G2G in circosdisplay characteristicsoff mostt elementslt arecustomizabledata-drivendatadrivenformatting rulessupport for datalayers

2D data in circos

2D data in circosboxscatterline

2D data in circostilestileshistogramheatmapschr2

non-linearnonlinear scalingglobal scaling – scaleoff eachh idideogramcan be adjustede g chr 1 drawn at 8xe.g.local scaling – anyregion can be locallyexpandedd d orcontractede.g.eg 100-150 Mb onchr1 expanded 5x

non-linearnonlinear scaling

circos in comparative genomicsmouse chr3mouse chr1human chr1

circos in comparative genomicschlamydia D fingerprint mapvschlamydia D sequence

circos in comparative genomicschlamydia L fingerprint mapvschlamydia D sequence

blast of regions of chr14 vs chr22alignments drawn asribbonsibbsinglealignment

circos is flexible

scircos art

visualizing quantitative informationvisualizing quantitative information martin krzywinski. outoutlineline best practices of graphical data design data-to-ink ratio cartjunk . reveal data at several levels of detail - broad overview and fine structure the visual display of quantitative information edward r tufte, 2001, 2nd ed.