Table 2.

Significant Predictors (and Their Sources) in the Regression Analyses of Human, Bat, and Ex Vivo Integrations of DNA Transposons.

Category	Predictor	Studied in Human, piggyBat and SB	Studied in Bat	Measure (transformation)a	Source	Description
DNA conformation	A-phased repeat	X	X	Content	Cer et al. (2011)	Runs of four or more As or Ts without the flexible TpA step (Rohs et al. 2009)
	Direct repeat	X	X	Content	Cer et al. (2011)	Two tracts of 10–50 nt separated by 0–5 nt that have the same composition
	G-quadruplex repeat	X	X	Count (log10) for human, content for bat	Cer et al. (2011)	Four blocks, with the same number of G bases (from 3 to 7), separated by 1–7 nt
	Inverted repeat	X	X	Content	Cer et al. (2011)	Two consecutive DNA sequences (10–100 nt long) separated by 0–100 nt that are palindromic on the same strand. They may fold back and generate double helices
	Mirror repeat	X	X	Count	Cer et al. (2011)	Two perfect repeats of 10–100 nt separated by 0–100 nt on the same strand
	Triplex motif	X		Count (log10)	Cer et al. (2011)	Similar to Mirror repeat but the repeat can contain only purines or pyrimidines on the same strand, the separation is at most 8 nt. These sequences can form three-stranded isoforms
	Z-DNA repeat	X	X	Content	Cer et al. (2011)	Five or more tandem repeats, each with alternating pyrimidine–purine dinucleotide motif in which the pattern YG is maintained on one of the DNA strands. These motifs can take the Z-DNA conformation
DNA sequence	Mononucleotide STR	X	X	Content	Genome-wide screen	All repeats of one nucleotide with length >9 bp
	Dinucleotide STR	X	X	Content	Genome-wide screen	All repeats of two nucleotide motifs with length >10 bp
	Trinucleotide STR	X	X	Content	Genome-wide screen	All repeats of three nucleotide motifs with length >12 bp
	Tetranucleotide STR	X	X	Content	Genome-wide screen	All repeats of four nucleotide motifs with length >16 bp
	TRF	X		Content (log10)		Repeats identified by the program Tandem Repeat Finder (Benson 1999)
	SINE	X	X	Count (log10) for human, content for bat	UCSC Genome Browser	Transposable element (Short INterspersed Elements)
	LINE	X	X	Count for human, content for bat	UCSC Genome Browser	Transposable element (Long INnterspersed Elements)
	L1 target sequence	X	X	Count	Cost et al. (2002)	Sequence associated to target primed reverse transcription which is characteristic of L1 and Alu mobilization
	Telomere hexamer sequence	X	X	Count	Morrish et al. (2007); Nergadze et al. (2007)	Sequence associated with DSB and repair by telomerases, and L1 retrotransposition as well
	CpG dinucleotides	X		Content	Genome-wide screen	Proportion of CpG dinucleotides in each window
	Gene or exon	X	X	Content	UCSC Genome Browser	All annotated genes or exons in these genomes
	Most conserved elements	X		Count	Siepel et al. (2005)	Regions of the genome that contain functional elements identified by comparative genomics
Expression and regulation	Chromatin accessibility	X		Count (sqrt)	UCSC Genome Browser	Regions of the genome digested by DNAse I reflecting active chromatin
	RNA polymerase II occupancy	X		Content	UCSC Genome Browser	Regions of the genome transcribed by RNA polymerase II
	DNA methylation	X		Count (sqrt)	Down et al. (2008)	Epigenetic modification that can modify the chromatin structure and regulate gene expression
	CpG islands	X	X	Content (sqrt) for human, count for bat	UCSC Genome Browser	Regions of the genome rich in GC close to promoters
Recombination	Recombination hotspots	X		Count (log10)	Myers et al. (2005)	Predicted recombination hotspots using SNP data
Position on the chromosome	Distance to telomere	X		Distance in bp (sqrt)	Genome-wide screen	Distance from the tip of the chromosome to each genome window defined here
	Distance to centromere	X		Distance in bp (sqrt)	Genome-wide screen	Distance from the centromere to each genome window defined here
Replication	Replication timing	X		Weighted average (sqrt)	Ryba et al. (2010)	Genome-wide microarray data measuring time in which replication occurs in hESC (human embryonic stem cells)

^aCount is the number of each feature in a particular window. Content is the fraction of a particular window that is occupied by a feature. Weighted average is used when several data intervals overlap within a window border.