Hi all,

I'm struggling with how to correctly specify my design formula within DESeq2 and any guidance would be much appreciated!

Overall objective: Contrast families at the biological replicate level while accounting for lane and index of a given technical replicate

I have 150 biological replicates. Groups of biological replicates correspond to a family. Some families have 1 biological rep and some have 5. Every biological replicate has three technical replicates.

Here would be an example of the colData

bio_rep                technical_rep  index  lane
Jones_BIOrep1            1            A        1
Jones_BIOrep1            2            A        2
Jones_BIOrep1            3            A        3
Jones_BIOrep2            4            B        3
Jones_BIOrep2            5            B        4
Jones_BIOrep2            6            B        1            
Martha_BIOrep1           7            C        1
Martha_BIOrep1           8            C        4
Martha_BIOrep1           9            C        3
Martha_BIOrep2           10           D        1
Martha_BIOrep2           11           D        4
Martha_BIOrep2           12           D        2
Martha_BIOrep3           13           E        1
Martha_BIOrep3           14           E        4
Martha_BIOrep3           15           E        2
....
....
Zach_BIOrep1             448           B        3
Zach_BIOrep1             449           B        4
Zach_BIOrep1             450           B        1

The basic design:

~ 0 + bio_rep

checks out and resultNames() returns the names of the 150 biological replicates

However I'm struggling adding in lane and index becuase not every techincal replicate is assigned every index or lane.

I've tried creating a model.matrix with:

~ index + index:bio_rep

and removing columns which have 0, but this did not seem to work.

I apologize if this is somewhere else, I haven't been able to find an example that I feel is similar.