Integration patterns in control versus MxB-expressing cells. (A) Number of integration sites counted in 1.25 kb bins (x-axis ticks) were plotted as percent of total from 30 kb upstream (negative x-axis value) to 30 kb downstream of TSSs. Blue and red lines, data from control and MxB-expressing HOS cells, respectively; The green plot, MRC values. The sites ~4 kb to 12 kb upstream of TSSs with significantly greater levels of integration than random in WT HOS cells in the vast majority of cases also mapped within RefSeq genes, which was attributed to the presence of internal promoters in the human genome . (B) Frequency of HIV-1 integration sites surrounding CpG islands. Line colorings are same as in panel A. See Additional file 3: Figure S3 for statistical analysis of panel A and B results. (C) Percent integration sites (y-axis) plotted against number of genes per Mb (x-axis) for infections conducted using control (blue line) and MxB-expressing (red line) HOS cells. Wilcox Rank-Sum test analysis of gene density targeting values (Table 1) yielded P values < 2.2 × 10302 for all comparisons (WT versus MRC, MxB versus MRC, and WT versus MxB). Fisher’s exact test was therefore conducted on the subset of integration sites that fell between 8 and 19 genes per Mb (bracket). Herein, WT and MxB-expressing cells harbored 34.8% and 35.4% of all integrations, respectively; the MRC value was 30.2%. The WT versus MRC comparison yielded P <2.2 × 10-308. MxB-expressing cells versus MRC yielded P =2.8 × 10-124 whereas P =0.01 was determined for control versus MxB-expressing cells. (D) Same as in panel C, except the following data from Ocwieja et al.  was analyzed: control siGL2, 7,140 integration sites; TNPO3 si4, 3,923 sites; RANBP2 si6, 2,114 sites. TNPO3, transportin 3. The panel D graph was smoothed using kernal estimation  due to relatively fewer numbers of integration sites.