Show Expression Changes from One Cluster

[drives28]

Hello,

I have the following design matrix (all columns are characters):
group time replicate ZN_D0_1 control 0 1 ZN_D0_2 control 0 2 ZN_D0_3 control 0 3 ZN_D1_1 control 1 1 ZN_D1_2 control 1 2 ZN_D3_1 control 3 1 ZN_D3_2 control 3 2 ZN_D3_3 control 3 3 ZN_D5_1 control 5 1 ZN_D5_2 control 5 2 ZN_D5_3 control 5 3 I23_D0_1 Trt1 0 1 I23_D0_2 Trt1 0 2 I23_D0_3 Trt1 0 3 I23_D1_1 Trt1 1 1 I23_D1_2 Trt1 1 2 I23_D1_3 Trt1 1 3 I23_D3_1 Trt1 3 1 I23_D3_2 Trt1 3 2 I23_D3_3 Trt1 3 3 I23_D5_1 Trt1 5 1 I23_D5_2 Trt1 5 2 I23_D5_3 Trt1 5 3 E33_D0_1 Trt2 0 1 E33_D0_2 Trt2 0 2 E33_D0_3 Trt2 0 3 E33_D1_1 Trt2 1 1 E33_D1_2 Trt2 1 2 E33_D1_3 Trt2 1 3 E33_D3_1 Trt2 3 1 E33_D3_2 Trt2 3 2 E33_D3_3 Trt2 3 3 E33_D5_1 Trt2 5 1 E33_D5_2 Trt2 5 2 E33_D5_3 Trt2 5 3

I have used DEGPatterns to obtain clusters for genes which show effects due to treatment (group) over time. I would like to show the boxplots for each group from a specific cluster over time. Similar to this individual's figure: https://github.com/lpantano/DEGreport/issues/35

I thought there was an issue with my design matrix. I have tried having the time column as a numeric and as a character. I have also tried to merge according to the replicate column but to no avail. I have looked at the vignette for DEGreport, but I could have missed something.

Any help would be appreciated, and I can provide more details if necessary.

[drives28]

If someone else is struggling with this, @Ipantano had suggested the colData(dds) ['time'] should be a factor with issue #42. I changed my metadata columns to all be factors as opposed to characters and this fixed my issue. I now see all clusters of genes per group and the expression profile each cluster of genes exhibits over time. Hope this can help someone in the future.

[lpantano]

Thank you! Yes, that is the common issue. I will leave this open so people can see it quickly. Thanks again for the follow up.

[drives28]

No problem. Do you know if the metadata being factors/characters/numeric affects the DESeq2 output, or if this is specific to degPatterns? I used the same metadata for both and my previous analyses were a mix of characters and numerics. I'm wondering if I need to redo these analyses.

[lpantano]

No, it should be fine. DESeq2 will convert to factor, something I should do to be honest, but I always forget...

[drives28]

Thank you so much for your help and discussion!