search

maqin2001 / qubic-r-package

Other
0 stars 0 forks source link

Step 2.optional. scmgraph (multiple dense graphing and merging) #5

Open maqin2001 opened 6 years ago

maqin2001 commented 6 years ago

update 04-10-2018

  1. read multiple bicluster results (files)
  2. do the same dense graph as "scgraph" for each file
  3. generate a zero matrix which col and row number depends on the total cell number (column) in the expression data
  4. Do linear superposition by adding each dense graph to the zero matrix based on column names.
  5. Remove columns and rows that are complete 0. (since it is a symmetry matrix, the same cell will be removed from column and row)
  6. output the single merged dense graph matrix.
zy26 commented 6 years ago

The name "scgraphmerge" can not describe multiple "scgraph" properly. We can change the parameter of "scgraph" and let it accept vector of bicluster results, or we can use "mscgraph" or "scmgraph" as the name of multiple "scgraph".

maqin2001 commented 6 years ago

Either one works for me. Thanks

Get Outlook for iOShttps://aka.ms/o0ukef

From: zy26 notifications@github.com Sent: Friday, April 13, 2018 8:23:24 PM To: maqin2001/qubic-r-package Cc: Qin Ma; Author Subject: Re: [maqin2001/qubic-r-package] 3. scgraphmerge (#5)

The name "scgraphmerge" can not describe multiple "scgraph" properly. We can change the parameter of "scgraph" and let it accept vector of bicluster results, or we can use "mscgraph" or "scmgraph" as the name of multiple "scgraph".

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/maqin2001/qubic-r-package/issues/5#issuecomment-381293830, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ABarDfLRHOOJ4j6mT9yGDhMBs7aL8jVfks5toU-MgaJpZM4TPN_4.

PegasusAM commented 6 years ago

We will use scmgraph for this step. Dr. Zhang, any updates on this step?

PegasusAM commented 6 years ago

Dr. Zhang, here is a new plan for scmgraph: instead of creating the zero matrix, we amend every uncovered cells in each graph loaded and simply sum all matrix then. so the steps would be:

  1. read list
  2. find missing colnames (if use a hashset)
  3. implant uncovered cells by row and colomn and assign all 0s.
  4. rearrange the matrix sequencially
  5. merge matrix.
  6. remove remaining all 0 columns and rows
  7. export