Date: Fri, 29 Mar 2024 17:01:46 +0100 (CET)
Message-ID: <297936151.2903.1711728106051@localhost>
Subject: Exported From Confluence
MIME-Version: 1.0
Content-Type: multipart/related;
boundary="----=_Part_2902_419489494.1711728106050"
------=_Part_2902_419489494.1711728106050
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Content-Location: file:///C:/exported.html
Dear Flux authors/maintainters/developpers,
We (I plus Nicolas Philiipe & Mika=C3=ABl Salson) have been using FluxS=
imulator quite a lot recently in order to create a fair benchmark for RNA-S=
eq mapping softwares, and we must say that it is a really good piece of sof=
tware!
However we have faced some difficulties in the design of our simulated libr=
aries :
The first difficulty concerns the error model that Flux uses and generalize=
s for longer reads. In our simulations we want to produce RNA-Seq data with=
longer reads, as the trend seems to go that way (200bp or greater). But wh=
en we do that, the scaled error model generates a lots of errors, and I can=
understand that. But I think that the errors rate will get lower as techno=
logy will evolves. What I am asking you is how could we tune your errors mo=
del, to make fewer mistakes (say 1%) on longer reads? I have seen that we c=
an generate custom errors models, but since that data we want to simulate d=
oes not already exists this is not a possible alternative. What are your th=
oughts about that?
The second difficulty concerns the polyA-tales simulations and the aligment=
s reported in the bed file. Some reads does have a part of the polyA-tale a=
nd a part that comes from the genome, but the alignement does not mention t=
he polyA-tale part. It is then difficult for us to assess mapping position =
of those reads. Have I missed something?
Thank you in advance for your answers,
Best regards,
J=C3=A9r=C3=B4me Audoux.
------=_Part_2902_419489494.1711728106050--