vidjil issueshttps://gitlab.inria.fr/vidjil/vidjil/-/issues2019-10-28T12:56:03+01:00https://gitlab.inria.fr/vidjil/vidjil/-/issues/4024Clones with different analyses at diagnosis and follow-up2019-10-28T12:56:03+01:00Mathieu GiraudClones with different analyses at diagnosis and follow-upUsers sometimes report sequences with different analyses at diagnosis and follow-up (link to old issue ?)
@meidanis has recently seen such sequences (could you share them?).
The main reason is probably #3970. See also #2726.
We have to ...Users sometimes report sequences with different analyses at diagnosis and follow-up (link to old issue ?)
@meidanis has recently seen such sequences (could you share them?).
The main reason is probably #3970. See also #2726.
We have to be sure and to investigate and/or document things.
cc @flothoni https://gitlab.inria.fr/vidjil/vidjil/-/issues/3964Running MiXCR through Vidjil2020-09-28T16:44:10+02:00Mathieu GiraudRunning MiXCR through VidjilReported by @meidanis :
> The first thing his analysis does is to combine R1 and R2. Shall I do
this as a pre-processing step? That was my inclination, but I'm not sure
the resulting file from "mixcr align" is a fastq file. Maybe it ...Reported by @meidanis :
> The first thing his analysis does is to combine R1 and R2. Shall I do
this as a pre-processing step? That was my inclination, but I'm not sure
the resulting file from "mixcr align" is a fastq file. Maybe it is
written in some other format. How can I continue the analysis then?
> Even if I continue the analysis calling mixcr again for the next step,
will it produce a `.vidjil` file? Does it know how to do that?
> Finally, is mixcr (the binary, executable) already installed in the
containers?
cc @flothoni @mikael-sThonier FlorianThonier Florian2019-12-20https://gitlab.inria.fr/vidjil/vidjil/-/issues/3925No spike normalization and disconnected lines2019-07-15T11:33:22+02:00Mathieu GiraudNo spike normalization and disconnected lines@meidanis has situations where the diagnosis point was not normalized (when there is no / a few spike-ins, this is the good way to go).
In this case, there were no lines between the clones in the ~"client-graph". He will send us the `.v...@meidanis has situations where the diagnosis point was not normalized (when there is no / a few spike-ins, this is the good way to go).
In this case, there were no lines between the clones in the ~"client-graph". He will send us the `.vidjil` files before and after normalization.
cc @flothonihttps://gitlab.inria.fr/vidjil/vidjil/-/issues/3663Bikeshedding normalisation2022-04-25T16:58:09+02:00Mathieu GiraudBikeshedding normalisationAprès !378 :
- libellé exact des menus
- icône
Je m'en occuperai la semaine prochaineAprès !378 :
- libellé exact des menus
- icône
Je m'en occuperai la semaine prochainehttps://gitlab.inria.fr/vidjil/vidjil/-/issues/3650How does the user see the normalized_reads field?2019-01-10T15:21:23+01:00Mathieu GiraudHow does the user see the normalized_reads field?Raised by @meidanis on #3645.
(We do not directly display `normalized_reads`, but ratios computed with this number, and depending on what is selected)
We could have something like `(17 reads (0.19%), 18.5 normalized (0.23%))` that coul...Raised by @meidanis on #3645.
(We do not directly display `normalized_reads`, but ratios computed with this number, and depending on what is selected)
We could have something like `(17 reads (0.19%), 18.5 normalized (0.23%))` that could be shown on the detailed information window, but we definitely need a short version.
It could be (17 reads is the actual number of reads, 0.23% is the ratio computed with `normalized_reads`):
- `17 reads (0.23%)`
- `17 reads (0.23%N)`
- `17 reads (0.23)`
Note that most of the people that will use this feature will probabibly always have normalized reads (even with #3648). It's probably better not to disturb them with too many information and to show them something simple.
cc @mikael\-s @flothoni https://gitlab.inria.fr/vidjil/vidjil/-/issues/1726Évolution clonale : changement de V (V switch), similarité VDJ-aware2022-06-17T12:39:29+02:00Vidjil TeamÉvolution clonale : changement de V (V switch), similarité VDJ-awareEuh ? Demander à Bruxelles ?
***
Je ne me souviens plus des détails, mais c'était venu à Zurich lors d'une discussion avec Marleen ~"BRU-Bruxelles" et Hélène ~"PAR-Debré". Les relancer avant, ou au pire en discuter lors du VW.
***
Disc...Euh ? Demander à Bruxelles ?
***
Je ne me souviens plus des détails, mais c'était venu à Zurich lors d'une discussion avec Marleen ~"BRU-Bruxelles" et Hélène ~"PAR-Debré". Les relancer avant, ou au pire en discuter lors du VW.
***
Discuté au VW. Mail de Cristina ~"PAR-Debré" :
> voici l'URL correspondant au patient que nous avons regardé tout à l'heure (locus IgH oligoclonal), passé sur le Miseq en V3 avec le kit LymphoTrack FR1 : http://app.vidjil.org/browser/?patient=1292&config=25. Il s'agit d'un VH4DH3JH6 et d'un VH6DH3JH6, partageant donc le même D-J.
***
Faire un calcul de similarité V/D/J-aware, capable de dire : "oh, ces séquences sont très similaires avec une certaine e-valeur car N2/J similaire"... Et le lancer par exemple via le browser comme "plot by similarity"
(Problème: sera dépendant du seuil de détection de D.)
En lien aussi avec modèle de proba d'une recombinaison (Bristol fait cela).
***
- soit un calcul de similarité directement sur la représentation sortie du FineSegmenter
- soit, plus précis, on fait cela directement en programmation dynamique, aligner deux séquences en connaissant leur AligBox, et avoir un score différent pour les N.
***
@magiraud @mikael-s