search

matsengrp / cft

Clonal family tree
5 stars 3 forks source link

CFT reformat for non-seed partitions #178

Closed lauradoepker closed 7 years ago

lauradoepker commented 7 years ago

Figured I should formally submit this, since we haven't written it down anywhere:

Duncan already ran the non-seed partis partitions for BF520 (LN-C and LN-D). file path: /fh/fast/matsen_e/processed-data/partis/laura-mb/v9/partitions/

For IgG, IgK, and IgL, we'd like CFT trees for the top 3 largest partitions (minadcl to 100 sequences is okay, I think) /fh/fast/matsen_e/processed-data/partis/laura-mb/v9/partitions/Hs-LN-D-5RACE-IgG-300k/ /fh/fast/matsen_e/processed-data/partis/laura-mb/v9/partitions/Hs-LN-D-5RACE-IgK-150k/ /fh/fast/matsen_e/processed-data/partis/laura-mb/v9/partitions/Hs-LN-D-5RACE-IgL-150k/

I think this means that @metasoarous has to go through the CFT code and made it non-seed specific, for these trees. Ultimately, we'd like CFT to let us click on a given patients, see all the seed trees, see Duncan's graphics for the non-seed partitioning (i.e. /fh/fast/matsen_e/processed-data/partis/laura-mb/v9/partitions/Hs-LN-D-5RACE-IgG-300k/plots/partitions/overall.html), and also see the top three biggest non-seed trees for each IgG, IgK, and IgL. ...But this doesn't need to be clean anytime soon. A temporary deployment will help us in the meantime.

lauradoepker commented 7 years ago

Here's a summary of what we're interested in for non-seeded BF520 partitions: 1) W1 and M9 separate timepoints 2) laura-mb and laura-mb-2 runs 3) 4 independent subsample runs each (i-sub-0 through 3) 4) G, K, and L libraries 5) 3 largest families from each of these libraries AND specifically the libraries that contain BF520.1-igh and -igk (this will be 16 additional trees, I think) =144 trees + 16 = 160 trees total

The rationale for the BF520.1-containing families is that we'd like to choose "representative" G and K sequences from these families and see if we still get BF520.1-like qualities (i.e. HIV binding and neutralization) from the resultant antibodies.

metasoarous commented 7 years ago

Reconsidering this as the data-only build of unseeded data necessary to get @lauranoges and her rotation student unblocked. Considering #213 to be the top level issue now.