pPGK1 quantseq behaviour

[DimmestP]

Unlike pRPS3 and pTSA1 constructs pPGK1 construct do not correlate well with qPCR data. I am currently trying to verify if this is part of the data or an artefact introduce from the analysis pipeline.

[DimmestP]

I have checked the multi-mapped reads of the pPGK1 mod_NNN and WT constructs. As expected the WT constructs have a higher rate of multi-mapping as they are identical to the genomic version of the RPS3 terminator. Nothing appears amiss with the mod constructs.

[DimmestP]

I have created a simulated set of 5 reads to test the pipeline. I have managed to get the pipeline to function properly and give me a read out. For some reason the 1st reads don't appear to land on the opposite side. I will check if I need to do the reverse complement to ensure the reads are mapped properly.

[DimmestP]

I have got the simulated counts running smoothly across the pipeline and creating cumulative counts graphs. I realised I was using the bedtools intersect function incorrectly. It created an entry for every time a read and a feature intersected. Therefore, one read could have multiple entries if it intersected with multiple features (up to 6 entries; terminator, primary transcript and motifs 1,2 and 3!). Now added the flag -u which ensures each read will only appear once if it overlaps multiple features.

[DimmestP]

I am still trying to understand why pPKG1 mod_NNN counts are low. I have checked whether it is mislabled (and might not actually be mod_NNN) but tried to align to mod_HTH and a got worst alignment. I also looked at the alignment on IGB and do not see any telltale signs of misalignment. The two replicates are very similar despite having varying sequencing depths. Also they both have high 70% to low 80% read alignment. Frustratingly, if I normalise all RNA-seq reads to WT instead of mod_NNN I get perfect correlation with qPCR reads.

[DimmestP]

Chatted with Weronika. Perhaps there is something odd with URA3 normalisation? Could URA3 be present in the genome fasta file? More or less reads could be being mapped their instead of the plasmid URA3? Alternatively we could pick 3 other genomic genes to normalise to (plus just to total reads).