Clarification on “-seq TA” Argument and Criteria for Identifying Essential Genes

[wzylin]

Dear TSAS Team,

I hope you are well. I have two questions regarding my recent analyses with TSAS:

1. Effect of the “-seq TA” Argument
   I performed analyses using the Mariner transposon method both with and without the “-seq TA” argument. In both cases, the output files were identical. Could you confirm whether this flag should affect the results for Mariner data, or is its behavior expected to be neutral in this context? From the .wig files generated, some read counts still shown up at non TA sites.

2. Selecting the Appropriate Adjusted p-Value for Essential-Gene Calls
   To identify essential genes, TSAS provides two adjusted p-value metrics:

Adj. p-value (proportions_insertions) — ~50 genes with p < 0.05

Adj. p-value (proportions_reads) — > 3 000 genes with p < 0.05

Which metric is recommended for determining essential genes in a typical Mariner dataset? Any guidance on best practices (or relevant documentation) would be greatly appreciated.

Thank you in advance for your assistance.

[srimam]

Hi there,

It's a little hard to answer these questions without any detail of the type of analysis you've done or what parameters you used. Can you attach the TSAS.log file so I can get some idea? In the meantime:

1. The for the "-seq" parameter, if you are doing a two-sample analysis this may not have a major impact as this normalization factor cancels itself out between the two samples (ie the total number of possible insertion sites is the same between both samples), so yes you should not see any effect. If you were doing a one sample analysis then you would see a major impact. The wig files don't filter out any hits, so yeah if there were some non-specific insertions they would show up there (that's part you why it isn't filtered so you user can make these observations). If this is a feature you would find useful (ie filtering out non-TA hits), let me know and I can include this in a future update.

2. As for then number of (conditional) essential genes and cutoffs to use, again without knowledge of parameters used for the run it's had to help. The numbers you provided suggest there might be a lot of PCR artifacts in one of the samples which wasn't adequately handled by the capping setting selected (given the large disparity between the insertion and read based metrics). If that's the case setting capping to 2 or 3 might be more appropriate. In general the proportions statistic will give you more power to detect essential genes as it take all reads into account, but can be prone in PCR bias if not appropriately handled.

Let me know if you have additional questions.

Thanks