The multiqc report highlighted potential issues with sample and/or library prep. A large proportion of reads were < 150bp read length (Illimina MiSeq) which seemed to result in negative 'Inner Distance'. Slight RNA degradation was also detected on the Gene Body Coverage graph. For the 'Latent' sample Jon noted that based on the complexity curve and the FastQC quality scores that there is a 'lot of incorrectly called bases along the length of the reads in the latent sample. Luckily, because this low quality is fairly uniform across the reads the alignment scores seem to be ok. So that sample is probably still good for expression analysis, but I wouldn’t trust it for any variant calling'. The multiqc report and the gene count matrix (which is very sparse) was sent to Walter, who agreed that they were having trouble with library prep and sequencing depth.
Processing has been delayed somewhat by the Nicol project. Will now start testing on this first dataset. Still expecting more files from Walter who says " I am expecting about 36 files with an average of 1M reads of about 150pb...I am worried about the depth. However, i think it is better to run it and see if there is enough resolution to detect DE. "
I met with Walter to get a update on the status of this project. After some delay last year in the lab he is now expecting to have the sequencing data in March this year. NB: for now this is bulk RNAseq on single cell as initially stated.
The experiment consists of FACS sorting of a) primary CD4 T-cells (from patients) and b) the Jarket CD4 T-cell line that have been infected with HIV. Two reporter genes are used to separate cells by FACS into three subsets: HIV-negative, HIV-latent and HIV-active. In addition the HIV-latent population were activated, again applying FACS sorting. The question Walter wants to answer is what differentiates latently-infected cells from cells that are actively infected (i.e. virions replicating) in terms of human gene expression in these cells.
Out existing RNAseq pipeline should be sufficient to process the expected Illumina MiSeq data however R scripts will have to be developed for downstream differential abundance testing.