Amend

[aspitaleri]

Hi all,
the TBAmend step require a lot of time in case of several inputs. I am wondering whether there an option in TBAmend by which adding new samples in a previous run would increase the speed without redoing the TBAmed for all samples, including the previous one. So basically, MTBSeq could just compare the new samples with the previous fasta to calculate the new matrix. I know that MTBseq pipeline require some restraints to make the TBAmend but in first approximation using the previous matrix could help in terms of speed. What do you think? Does it need some perl-hacking?
Best

[TaKohl]

Hello Andrea, really good point with TBAmend, and an interesting idea to shave off calculation time by using the previous calculation as a basis. Problem is, I imagine this to be quite problematic to implement as a generic solution. Having a recalculated joint analysis with just new samples added is a very specific case. We are currently looking into options on how to improve a joint analysis in general, e.g. by splitting up the calculation into several parallel processes or switching to consensus. However, if you can implement a solution along the lines you described, that would be interesting to see.