Hello,
I have repeated sequencing of some samples because we wanted higher depth (at least 50 million)
For most replicates, both batches passed the QC, but we didnt repeat any libraries, just resequenced.
For some samples, just batch 1 was good enough, for some, just batch 2.
But I can clearly see a batch effect even for technical replicates. Of course, technical replicates are spread accross my different biological groups
Would it ok to do CombatSeq in all samples and then merging the technical replicates (summing counts) after CombatSeq? I could find the answer for Combat, but not CombatSeq
In my head, having technical replicates in both batches would help CombatSeq find specific batch errors and not overcorrect anything.
Example of my design:
| Sample_ID_batch |
Sample_ID |
Batch |
| 1_batch1 |
1 |
batch1 |
| 1_batch2 |
1 |
batch2 |
| 2_batch1 |
2 |
batch1 |
| 2_batch2 |
2 |
batch2 |
| 3_batch1 |
3 |
batch1 |
| 3_batch2 |
3 |
batch2 |
| 4_batch1 |
4 |
batch1 |
| 4_batch2 |
4 |
batch2 |
| 5_batch1 |
5 |
batch1 |
| 5_batch2 |
5 |
batch2 |
| 6_batch1 |
6 |
batch1 |
| 6_batch2 |
6 |
batch2 |
| 7_batch1 |
7 |
batch1 |
| 8_batch1 |
8 |
batch1 |
| 8_batch2 |
8 |
batch2 |
| 9_batch1 |
9 |
batch1 |
| 9_batch2 |
9 |
batch2 |
| 10_batch1 |
10 |
batch1 |
| 10_batch2 |
10 |
batch2 |
| 12_batch1 |
12 |
batch1 |
| 12_batch2 |
12 |
batch2 |
| 14_batch1 |
14 |
batch1 |
| 14_batch2 |
14 |
batch2 |
| 15_batch1 |
15 |
batch1 |
| 16_batch1 |
16 |
batch1 |
| 17_batch1 |
17 |
batch1 |
| 20_batch1 |
20 |
batch1 |
| 21_batch2 |
21 |
batch2 |
| 22_batch1 |
22 |
batch1 |
| 22_batch2 |
22 |
batch2 |
| 23_batch1 |
23 |
batch1 |
| 24_batch1 |
24 |
batch1 |
| 24_batch2 |
24 |
batch2 |
| 25_batch1 |
25 |
batch1 |
| 25_batch2 |
25 |
batch2 |
| 26_batch1 |
26 |
batch1 |
| 26_batch2 |
26 |
batch2 |
| 28_batch1 |
28 |
batch1 |
| 28_batch2 |
28 |
batch2 |
| 29_batch1 |
29 |
batch1 |
| 29_batch2 |
29 |
batch2 |
| 31_batch1 |
31 |
batch1 |
| 32_batch1 |
32 |
batch1 |
| 35_batch1 |
35 |
batch1 |
| 36_batch1 |
36 |
batch1 |
| 36_batch2 |
36 |
batch2 |
| 37_batch1 |
37 |
batch1 |
| 38_batch1 |
38 |
batch1 |
| 38_batch2 |
38 |
batch2 |
| 39_batch1 |
39 |
batch1 |
| 39_batch2 |
39 |
batch2 |
| 41_batch1 |
41 |
batch1 |
| 41_batch2 |
41 |
batch2 |
| 42_batch1 |
42 |
batch1 |
| 42_batch2 |
42 |
batch2 |
| 43_batch1 |
43 |
batch1 |
| 44_batch1 |
44 |
batch1 |
| 45_batch1 |
45 |
batch1 |
| 46_batch1 |
46 |
batch1 |
| 46_batch2 |
46 |
batch2 |
| 47_batch1 |
47 |
batch1 |
Hello,
I have repeated sequencing of some samples because we wanted higher depth (at least 50 million)
For most replicates, both batches passed the QC, but we didnt repeat any libraries, just resequenced.
For some samples, just batch 1 was good enough, for some, just batch 2.
But I can clearly see a batch effect even for technical replicates. Of course, technical replicates are spread accross my different biological groups
Would it ok to do CombatSeq in all samples and then merging the technical replicates (summing counts) after CombatSeq? I could find the answer for Combat, but not CombatSeq
In my head, having technical replicates in both batches would help CombatSeq find specific batch errors and not overcorrect anything.
Example of my design: