In addition, Tat or other viral many proteins may affect cellular transcription. To look at transcription and latency, we ran a logistic regression of silentinducible status on a quartic func tion of RNA expression, as determined by RNA Seq reads within 5,000 bases in Jurkat cells for the Jurkat sample or CD4 T cells for the remaining samples, interacted with indicator variables encoding cell culture model. There appears to be little agreement between samples. The Resting CD4 and Active CD4 datasets show an enrichment in silent proviruses in regions with low gene expression. The other three studies show the opposite or no relationship for low expression regions. The two samples showing increased silence in areas of low expres sion are from a study that did not check whether inactive viruses could be acti vated.

One possible explanation is that regions with low gene transcription Inhibitors,Modulators,Libraries may harbor proviruses that are not eas ily activated, though some other discrepancy between in vitro systems could also explain the difference. Both the Jurkat and Active CD4 samples appear to increase in latency with increasing expression while the remaining three studies did not show a strong trend. Orientation bias Shan et al. reported that inducible proviruses were oriented in the same strand as the host cell genes into which they had integrated more often than chance. This orientation bias was still reproduced after our reprocess ing of the Bcl 2 transduced CD4 sample from Shan et al. However, the proportion of provirus oriented in the same strand as host genes did not differ significantly from 50% in the other samples.

Perhaps orienta tion bias and transcriptional Inhibitors,Modulators,Libraries interference are especially sensitive to parameters of the model system. Gene deserts Lewinski et al. reported increased latency in gene deserts. In Inhibitors,Modulators,Libraries the collected data, integration outside known genes was associated with latency. Inhibitors,Modulators,Libraries This seemed to largely be driven Inhibitors,Modulators,Libraries by the Active CD4 and Resting CD4 samples with significant associ ation found individually in only those two samples and no significant association observed in the other three samples. Looking only at integra tion sites outside genes, silent sites in the Resting CD4 sample had a mean distance to the nearest gene 2. 5 times greater than that of expressed sites. The Active CD4 sample had a small difference that did not survive Bonferroni correction.

Lewinski et al. also reported decreased latency near CpG islands and reasoned this was tied to the increased latency in gene deserts. In the Resting CD4 sample, silent sites were on average further from CpG islands than expressed sites, but there was no significant relation ship between silentinducible status and log distance to CpG island after Bonferroni correction if sellckchem the integration sites location inside or outside of a gene was accounted for first.