kaizhang
commented
7 years ago @npklein It is absolutely possible to rerun only certain steps. But this feature hasn't been documented. To rerun selected steps:
- Look at the workflow and decide which steps you want to rerun:
taiji view. The output is DOT code. You can usedotprogram to convert it to a graph or just open a text editor to view it. This is an example for the latest version oftaiji: https://github.com/kaizhang/Taiji/blob/master/Taiji.png. - Delete cache from the database using:
taiji rm [STEP_NAME]. For example,taiji rm RNA_alignment_prepare. In your case, you need to delete all steps related to RNA-seq. - Rerun the program. If you want to focus on particular steps, you can use:
taiji run --config config.yml --select RNA_alignment_prepare,RNA_alignment. This will only execute the specified steps and their dependencies.
Back to your problem, the program did not analyze the RNA-seq data because BAM file is currently not supported for RNA-seq analysis (see this table). You have two options:
- Use FASTQ as input.
- Analyze the BAM file by yourself and use the gene quantification as input.
If you decide to change input, you need to rerun "Initialization": taiji rm Initialization && taiji run --config config.yml --select Initialization.
npklein
Hi @kaizhang, after fixing the BAM chromosoom names the rest of the pipeline ran without an error, however the Network and Rank files are empty. The RNA_Seq output dir is also empty, so I'm thinking it might be because the input.yml RNAseq part is not done correct, but it looks like the example input file: https://gist.github.com/npklein/dd3acaad067fbf96ba03e46dd7d97c9a.
Also, in the logs it does say that it is doing something with the RNAseq
Also, is there a way to rerun only part of the pipeline? I can remove
sciflow.db, but then it also reruns ATAC-seq MarkDuplicates and peak calling, which takes quite long. Would like to test RNAseq only part of the pipeline (if that is the problem)