NEWS

---

SynExtend 1.17.7

• ExoLabel is much much faster and does a better job cleaning up when aborted early
• ExoLabel now has fewer arguments
• Updates to man pages

SynExtend 1.17.6

• Lots of internal improvements to ExoLabel to increase computational speed and decrease disk usage.
• ExoLabel will no longer crash if given relative paths.
• Adds more internal error checking to prevent some rare bugs.
• Updates man pages to reflect new changes.
• Updates EstimateExoLabel to reflect new changes.

SynExtend 1.17.5

• ExoLabel will no longer brick R during sorts on large files.
• ExoLabel reports more progress during some lengthy processing sections when verbose=TRUE
• Known issue: "Copying source file" step is still non-interruptable, will be fixed in a later update

SynExtend 1.17.4

• ExoLabel now allows an inflation argument to control application of inflation

SynExtend 1.17.3

• predict.EvoWeaver now supports returning p-values separately from raw score for some algorithms.
• OpenMP implementation for EvoWeaver algorithms that support it has been fixed

SynExtend 1.17.2

• RandForest function added to train random forest models
• Associated man pages for RandForest and DecisionTree objects
• New methods for DecisionTree objects to plot and coerce to dendrogram
• Small bugfix to subset.dendrogram

SynExtend 1.17.1

• Major updates to EvoWeaver:
  • predict.EvoWeaver now returns a data.frame by default
  • Method arguments are updated to match their names in the associated EvoWeaver manuscript
  • Above changes have propagated to documentation files
  • New Phylogenetic Profiling methods with improved accuracy
  • New meta-methods PhylogeneticProfiling, PhylogeneticStructure, GeneOrganization, SequenceLevel for predict.EvoWeaver
  • New pre-trained Ensemble models have been included
• Updates to ExoLabel for better status printing

SynExtend 1.16.0

• Bioc 3.19 release

SynExtend 1.15.3

• Addition of FastLabelOOM function to find communities in graphs/networks on disk space.

SynExtend 1.15.2

• Addition of PrepareSeqs function, beginning the process of deprecating PairSummaries in favor of more cohesive and user friendly functions.

SynExtend 1.15.1

• Fixes bug in JRF distance causing scores to be higher than expected.

SynExtend 1.13.8

• Updates to all EvoWeaver documentation files
• Fixed small bug in PhyloDistance causing Method='JRF' to return similarity rather than the distance
• Fixed small bug in TreeDistance.EvoWeaver resulting in an inconsistent calculation of score when using TreeMethods='JRF'

SynExtend 1.13.7

• Small fixes

SynExtend 1.13.6

• ProtWeaver and ProtWeb have been renamed to EvoWeaver and EvoWeb, respectively
• New sequence level method for EvoWeaver
• Various small internal updates to EvoWeaver

SynExtend 1.13.5

• Minor changes to SelectByK and vignette

SynExtend 1.13.4

• New predictor PAPV.ProtWeaver to calculate p-values for presence/absence profiles.
• ContextTree now uses MirrorTree with species tree correction and p/a overlap correction
• Updates to documentation

SynExtend 1.13.3

• predict.ProtWeaver now supports multiple algorithms at once (ex. predict(ew, Method=c("Jaccard", "Hamming")))
• Documentation for ProtWeaver and associated methods has been updated to match recent updates.

SynExtend 1.13.2

• FastQFromSRR function added as a convenience wrapper for the SRAtoolkit function fastq-dump.

SynExtend 1.13.1

• SuperTree now works directly with dist objects, providing better performance and scaling
• Updates to simMat objects
  • No longer throw a warning when initialized in RStudio
  • Formatting is cleaner and supports larger object names
• Updates to NVDC.ProtWeaver
  • Now supports amino acid sequences using the DNAseqs=FALSE argument
  • Now calculates a p-value-weighted score
• Adds MakeBlastDb function to create a BLAST database from R, plus associated documentation updates
• Smaller fixes to some ProtWeaver methods
  • predict.ProtWeaver no longer returns using invisible (this was annoying and unneccessary)
  • APC correction for MutualInformation.ProtWeaver removed to allow for parallelization
  • MirrorTree.ProtWeaver now works correctly with MTCorrection="speciestree"
  • CorrGL.ProtWeaver now uses Fisher's Exact Test for p-values rather than the R value of spearman correlation
• Many internal performance improvements
  • ProtWeaver almost entirely uses dist objects rather than matrix, saving significantly on memory
  • faster Cophenetic function implemented internally
  • Copied internal .Call('cophenetic') from DECIPHER to SynExtend to avoid potential namespace issues
  • Small fixes to remove some notes from BiocCheck::BiocCheck()
• Variety of small updates to pass BiocCheck

SynExtend 1.12.0

• Official Bioconductor 3.17 release (even with SynExtend 1.11.8)

SynExtend 1.11.8

• Fixes various small bugs in MoransI
• Adds some multiprocessing support (more will be added in the future)
• Slight rework to species trees and their interaction with ProtWeaver objects
  • ProtWeaver has new attribute speciesTree, can be initialized with a dendrogram object
  • New method SpeciesTree to get species tree from a ProtWeaver object (or compute one, if it doesn't exist)
• Various internal improvements for Bioc style consistency
• Various documentation updates

SynExtend 1.11.7

• Adds new optimized dendrapply implementation (overloads stats::dendrapply)
• AddsHungarianAlgorithm for optimal solving of the linear assignment problem (O(n^3) complexity)
• Adds new C code for fast computation of Pearson's R and p-value
• Adds new Ancestral.ProtWeaver algorithm for calculating coevolution from correlated residue changes
• Supporting code and documentation for Ancestral.ProtWeaver
• Other new internal methods
• Various updates and optimizations to internal methods and documentations
• Updates GRF method to be called CI (for Clustering Information Distance)
• Method="CI" in PhyloDistance now calculates an approximate p-value using simulated data from Smith (2020)

SynExtend 1.11.6

• Adds new Residue method NVDT using gene sequence Natural Vector with Dinucleotide and Trinucleotide frequency
• Adds some new C methods to speed up calculation of NVDT
• Fixes .Call() not using PACKAGE="SynExtend"
• Updates to documentation

SynExtend 1.11.5

• Adds new colocalization algorithm ColocMoran, uses Coloc with MoransI to correct for phylogenetic signal
• Adds new colocalization algorithm TranscripMI, uses mutual information of transcriptional direction
• Adds new corrections/checks to allow for transcriptional direction to be in labels
• Various bugfixes to support new four number labelling scheme
• Various documentation updates
• Adds new function MoransI to calculate Moran's I for a set of spatially distributed signals

SynExtend 1.11.4

• Internal code refactor
• ShuffleC now supports reproducibility using R's set.seed
• ShuffleC now support sampling with replacement, performance is around 2.25x faster than sample

SynExtend 1.11.3

• Internal bugfixes for JRF Distance--previous commit was incorrectly calculating values
• Adds new TreeDistance predictor for ProtWeaver, incorporating all tree distance metrics; these metrics are bundled due to some backend optimizations that improve performance
• Bugfixes for PhyloDistance
• Adds Random Projection for MirrorTree predictor to solve memory problems and increase accuracy
• New internal random number generator using xorshift, significantly faster than sample()
• HammingGL changed to CorrGL, now uses Pearson's R weighted by p-value
• Refactors internal predictors to reduce size of codebase and remove redundancies
• Internal ShuffleC function to replicate sample functionality with 2-6x speedup
• Method GainLoss now uses bootstrapping to estimate a p-value
• Updates to documentation files

SynExtend 1.11.2

• Adds KF Distance for trees
• Adds Jaccard Robinson Foulds Distance for trees
• Reworks tree distances into PhyloDistance function
• Numerous new documentation pages
• Updates internal functions to use rapply instead of dendrapply to avoid stack overflow issues due to R recursion

SynExtend 1.11.1

• Minor bugfix to RF distance
• updates gitignore for workflows

SynExtend 1.10.1

• Memory leak bugfix

SynExtend 1.9.21

• Adds new RFDist function to calculate Robinson-Foulds Distance
• Adds normalization for GeneralizedRF to make the distance between 0 and 1
• Minor bugfixes
• Documentation for new functions

SynExtend 1.9.20

• Adds new GeneralizedRF function to calculated information-theoretic Generalized Robinson-Foulds distance between two dendrograms.
• Documentation for new function
• New ProtWeaver predictor based off of GeneralizedRF metric
• New internal C source code for GeneralizedRF

SynExtend 1.9.19

• Adds new DPhyloStatistic function to calculate the D-statistic for a binary state against a phylogeny following Fritz and Purvis (2009).
• Documentation for new function
• new internal C source code for DPhyloStatistic
• new internal C source code for random utility functions, currently only has functions to generate random numbers

SynExtend 1.9.18

• Various internal improvements to presence/absence profile methods

SynExtend 1.9.17

• Adds new prediction algorithm GainLoss
• Adds new internal C implementation of dendrograms, significantly faster than R dendrograms
• ProtWeaver methods Behdenna and GainLoss can now infer a species tree when possible
• Updates Jaccard and Hamming methods to use C implementations for distance calculation
• Adds HammingGL method to calculate Hamming distance of gain/loss events
• Minor bugfixes to ProtWeaver methods relating to subsetting
• Updates to various man pages

SynExtend 1.9.16

• Removes flatdendrapply, function was already included in SynExtend
• minor bugfixes to ProtWeaver

SynExtend 1.9.15

• Edits to SelectByK, function can work as intended, but is still too conservative at false positive removal.

SynExtend 1.9.14

• Adds new function flatdendrapply for more options to apply functions to dendrograms. Function is used in SuperTree.
• Adds new function SuperTree to construct a species tree from a set of gene trees.
• Adds new dataset SuperTreeEx for SuperTree and flatdendrapply examples.

SynExtend 1.9.13

• SelectByK function argument ClusterSelect switched to ClusterScalar. Cluster number selection now performed by fitting sum of total within cluster sum of squares to a right hyperbola and taking the ceiling of the half-max. Scalar allows a user to pick different tolerances to select more, or less clusters. Plotting behavior updated.

SynExtend 1.9.12

• simMat class now supports empty indexing (s[])
• simMat class now supports logical accession (s[c(T,F,T),])

SynExtend 1.9.11

• Added the function SelectByK that allows for quick removal of false positive predicted pairs based on a relatively simple k-means approach. Function is currently designed for use on the single genome-to-genome pairwise comparison, and not on an all-vs-all many genomes scale, though it may provide acceptable results on that scale.

SynExtend 1.9.10

• New simMat class for dist-like similarity matrices that can be manipulated like base matrices
• Major update to ProtWeaver internals
• All internal calls use simMat objects whenever possible to decrease memory footprint
  • Note ContextTree and ProfDCA require matrices internally
• ProtWeb objects now inherit from simMat
  • ProtWeb.show and ProtWeb.print now display predictions in a more natural way
  • GetProtWebData() deprecated; ProtWeb now inherits as.matrix.simMat and as.data.frame.simMat
• New documentation pages for simMat class
• GetProtWebData documentation page reworked into ProtWeb documentation file.
• Fixes new bug in Method='Hamming' introduced in SynExtend 1.9.9

SynExtend 1.9.9

• Fixes minor bug in Method='Hamming'
• Moves some code around

SynExtend 1.9.8

• Major refactor to file structure of ProtWeaver to make individual files more manageable
• Adds new documentation files for individual prediction streams of predict.ProtWeaver

SynExtend 1.9.7

• BlockReconciliation now returns a an object of class PairSummaries.

SynExtend 1.9.6

• Fixes an error where warnings were mistakenly output to the user

SynExtend 1.9.5

• Moves platform-specific files in src/ (originally added by mistake)

SynExtend 1.9.4

• Lots of bugfixes to ResidueMI.ProtWeaver
• predict.ProtWeaver now correctly labels rows/columns with gene names, not numbers
• predict.ProtWeaver now correctly handles Subset arguments
• predict.ProtWeaver(..., Subset=3) will correctly predict for all pairs involving gene 3 (or for any gene x, as long as Subset is a length 1 character or integer vector).

SynExtend 1.9.2

• Adds residue MI method to ProtWeaver
• Various bugfixes for ProtWeaver

SynExtend 1.7.14

• Various improvements for GenRearrScen, improves consistency and output formatting
• Major bugfix for ProtWeaver methods using dendrogram objects
• ProtWeaver now correctly guards against non-bifurcating dendrograms in methods that expect it

SynExtend 1.7.13

• Introduces new ProtWeaver class to predict functional association of genes from COGs or gene trees. This implements many algorithms commonly used in the literature, such as MirrorTree and Inverse Potts Models.
• predict(ProtWeaverObject) returns a ProtWeb class with information on predicted associations.
• Adds BlastSeqs to run BLAST queries on sequences stored as an XStringSet or FASTA file.

SynExtend 1.7.12

• Updates to ExtractBy function. Methods and inputs simplified and adjusted, and significant improvements to speed.

SynExtend 1.7.11

• Updated NucleotideOverlaps to now correctly registers hits in genes with a large degree of overlap with the immediately preceding gene.
• Fixed aberrant behavior in BlockExpansion where contigs with zero features could cause an error in expansion attempts.

SynExtend 1.7.10

• BlockReconciliation now allows for setting either block size or mean PID for reconciliation precedence.

SynExtend 1.7.9

• Added retention thresholds to BlockReconciliation.

SynExtend 1.7.8

• BlockExpansion cases corrected for zero added rows.

SynExtend 1.7.7

• Improvements to BlockExpansion and BlockReconciliation functions.

SynExtend 1.7.5

• Began integration of DECIPHER's ScoreAlignment function.

SynExtend 1.7.4

• Fixed a bug in PairSummaries function.

SynExtend 1.7.3

• Added BlockExpansion function.

SynExtend 1.7.2

• Adjustment in how PairSummaries handles default translation tables and GFF derived gene calls.

SynExtend 1.7.1

• Large changes under the hood to PairSummaries.
• Failure to accurately assign neighbors in some cases should now be fixed.
• Extraction of genomic features is now faster.
• OffSetsAllowed argument now defaults to FALSE. This argument may be dropped in the future in favor of a more complex function post-summary.
• Small edits to SequenceSimilarity

SynExtend 1.5.4

• Added the function SubSetPairs that allows for easy trimming of predicted pairs based on conflicting predictions and / or prediction statistics.
• Added the function EstimageGenomeRearrangements that generates rearrangement scenarios of large scale genomic events using the double cut and join model.

SynExtend 1.5.3

• Added the function SequenceSimilarity and made improvements to runtime in DisjointSet.

SynExtend 1.4.1

• Fixed a small bug in consensus scores in PairSummaries where features facing on different strands had their score computed incorrectly.

SynExtend 1.3.15

• Changes to concensus score in PairSummaries.

SynExtend 1.3.14

• Major changes to the PairSummaries function and minor changes to NucleotideOverlaps, ExtractBy, and FindSets. Adjustments to the model that PairSummaries calls on to predict PIDs.

SynExtend 1.3.13

• ExtractBy function has been added. Allows extraction of feature sequences into XStingSets organized by the a PairSummaries object or the single linkage clusters implied by pairings within the PairSummaries objects.
• DisjointSet function added to extract single linkage clusters from a PairSummaries object.

SynExtend 1.3.12

• PairSummaries now computes 4-mer distance between predicted pairs.

SynExtend 1.3.11

• PairSummaries now returns a column titled Adjacent that provides the number of directly adjacent neighbor pairs to a predicted pair. Gap filling code adjusted.
• The function FindSets has been added and performs single linkage clustering on a pairs list as represented by vectors of integers using the Union-Find algorithm. Long term this function will have a larger wrapper function for user ease of access but will remain exposed.

SynExtend 1.3.10

• NucleotideOverlap now passes it's GeneCalls object forward, allowing PairSummaries to forego inclusion of that object as an argument.

SynExtend 1.3.9

• Minor vignette and suggested package changes.

SynExtend 1.3.8

• PairSummaries now allows users to fill in specific matching gaps in blocks of predicted pairs with the arguments AllowGaps and OffSetsAllowed.

SynExtend 1.3.7

• Adjustments to progress bars in both PairSummaries and NucleotideOverlap.
• PID prediction models in PairSummaries adjusted.

SynExtend 1.3.6

• Contig name matching has been implemented. Scheme expects users to follow NCBI contig naming and gff formats, accepting contig names from gffs directly, and removing the first whitespace and everything thereafter from FASTA headers. Contig name matching can be disabled if users wish, using the argument AcceptContigNames, but ensuring that the correct contigs in GeneCalls objects are matched to the appropriate contigs in Synteny objects are then the user's responsibility.

SynExtend 1.3.5

• PairSummaries now translates sequences based on transl_table attributes provided by gene calls
• PairSummaries now uses a generic model for predicting PID
• gffToDataFrame now parses out the transl_table attribute

SynExtend 1.3.2

• Minor changes to NucleotideOverlap
• Major changes to PairSummaries - can now take in objects of class Genes build by the DECIPHER function FindGenes()

SynExtend 0.99.30

• Vignette and help files edited for clarity

SynExtend 0.99.1

• SynExtend submitted to Bioconductor
• Added function gffToDataFrame
• Added function NucleotideOverlap
• Added function PairSummaries