NEWS

---

sbivar 0.99.35

• No more partial matching of method names in example code
• SpatialExperiment is now in 'Imports' as key Bioconductor dependency
• Add tests for lists of SpatialExperiment objects and with sample identifiers sample_id_x
• Dispatching versus X and Y
• Define an S4 results class

sbivar 0.99.31

• In BiocParallel, set stop.on.error as default to skip over failed GAM fits

sbivar 0.99.30

• Implemented some goodpractice suggestions
• Remove dependence on sp in favour of sf

sbivar 0.99.29

• Catch fit errors for GAMMs and fit regular GAMMs instead

sbivar 0.99.28

• More flexibility and better defaults for residual correlation structures

sbivar 0.99.27

• Switch to anndataR package rather than anndata and zellkonverter for reading in h5ad files
• Fix minor bug in plotGAMs

sbivar 0.99.26

• Remove GP smooth, it is too concurve with the other smooth after all
• Add GP error structure with mgcv::gamm instead, as an option

sbivar 0.99.25

• Remove unused offsets argument

sbivar 0.99.24

• Citation file and reference to preprint

sbivar 0.99.23

• Bug in plotSpotSums

sbivar 0.99.22

• Extend troubleshooting section in vignette
• scaleBySampleSums default FALSE in plotPairSingle
• New function for plotting the spatial distribution of the spot-wise sums

sbivar 0.99.21

• Round to significant digits in writeSbivarToXlsx()

sbivar 0.99.20

• Bug fixes when only one feature pair is tested

sbivar 0.99.19

• Estimate variograms only for features to be tested

sbivar 0.99.18

• Switch to lapply for buildGams

sbivar 0.99.17

• Added r-universe github build
• Switch to tempfile for illustrating writeSbivarToXlsx()

sbivar 0.99.16

• Provide featuresX and featuresY to sbivarMulti

sbivar 0.99.15

• Scale contributions to correlation by maximum absolute value in plotGAMs

sbivar 0.99.14

• Keep finite correlation estimates

sbivar 0.99.13

• Fix naming bug in extractResultsMulti

sbivar 0.99.12

• Add option to scale point sizes to library sizes in case of normalization or offset modelling

sbivar 0.99.11

• Add pseudocount for gamma family in plotGAM

sbivar 0.99.10

• Switch to SnowParam for multithreading on windows

• No more multithreading in vignette building on windows

sbivar 0.99.9

• Call gtsat:variogram and gstat::vgm explicitly

sbivar 0.99.8

• Use bpworkers(bpparam()) to retreive number of registered cores rahter than bpparam()$workers

sbivar 0.99.7

• Rearrange matrix multiplications for GAMs to save memory
• C++ speed-ups for GPs and Moran's I
• Fixing some potential issues in fitLinModels and extractResultsMulti
• Add Makevars for C++ Windows installation

sbivar 0.99.6

• Switch to common names for metabolites in Vicari data
• Switch off GP smoother in GAMs as default
• In plotGAMsTopResults, use provided families as default

sbivar 0.99.5

• Fix bugs in plotGAM
• Add feature names as columns in results

sbivar 0.99.4

• Fix bugs in testGAM
• Update Github readme

sbivar 0.99.3

• Include Gaussian random field in GAM fitting to account for stochastic neighbour resemblance

sbivar 0.99.2

• When Gamma fit fails, attempt negative binomial fit

sbivar 0.99.1

• Preparing submission to Bioconductor

sbivar 0.2.1

• Modify weighting parameters for Moran's I
• Add pseudocount for GAMs with gamma distribution
• Stricter row and column name checking

sbivar 0.2.0

• Implement bivariate Moran’s I for single images
• Disallow sample matching and correlation for disjoint coordinate sets
• Scale by maximum values for bivariate Moran’s I

sbivar 0.1.4

• Enforce x-y column names for the coordinates in sbivarMulti too

sbivar 0.1.3

• Make the package more verbose, printing progress on fitting and testing

sbivar 0.1.2

• Enforce valid names on data matrices with giveValidNames()

sbivar 0.1.0

• Package creation and first upload to Github