Question: Flow issues? (no temp SAM/BAM to start run?)
1
gravatar for dragon28
12 months ago by
dragon2830
dragon2830 wrote:

Hi, Trying to run a workflow. But cannot due to the insistence of cuffdiff on existence of SAM/BAM file(s) (which are not generated before the workflow is run).

This really seems to limit the utility of workflows. Is there any way to get past this (p.s. in prior runs of this workflow, this was not experience).

ADD COMMENTlink modified 11 months ago • written 12 months ago by dragon2830

First, rerun again now. The server is still undergoing updates/changes and this could be related.

If it still is an issue, could you explain more or share a link to the workflow and a test history with the inputs you use. I am wondering if the objects in the workflow are configured/linked correctly. In particular, I'd like to check if "Input" selections are included in the workflow (required - found at the top of the tool panel in the workflow editor - click the appropriate type to add these to the editor and link to the downstream tool(s)).

If there is a workflow editor issue, this will help to track that down as well. Plus we will be able to let you know if this is related to current server upgrades. Details/status about those here: https://biostar.usegalaxy.org/p/20733/#20854

Post here if you do not mind the history/workflow being accessible to all. Otherwise, a direct email with the links can be sent to galaxy-bugs@lists.galaxyproject.org (private list, internal to the team running the http://usegalaxy.org server). Please include a link to this post and send from your registered Galaxy account, if you choose to send an email. If there is a history that contains errors from the partial workflow run, that can be shared too (direct bug report). Please leave datasets (inputs/error jobs) undeleted.

How to share any Galaxy object with a link: https://wiki.galaxyproject.org/Learn/Share

Thanks! Jen, Galaxy team

ADD REPLYlink modified 12 months ago • written 12 months ago by Jennifer Hillman Jackson23k

I also ran into an issue with workflow but I am not sure if it is related to the issue mentioned here.

Yesterday I ran some workflows that had previously ran smoothly. Here is a link to the workflow:

https://usegalaxy.org/u/sushivision/w/mm9-mf10-noref-161207

Basically the main work flow goes like this: (1)input --> (2)mapped with bowtie --> (3)converted to bam --> (4)MACS --> (5)converted from wig to bigwig

The workflow also branches out with with (1)input -> FastQC and (2)bowtie (sam) -> (invisible internal BAM conversion) -> Flagstat

However this time the workflow only proceeded with the initial steps, i.e (1), (2) and (3) and FastQC. Flagstat and MACS (along with its downstream conversion) did not run --- did not appear in the history at all. (When I initiated the workflow all processes invoked were present, however.) I had to manually run the downstream steps myself, which offset the advantages of using workflow.

I hope that this is a temporary issue related to the move and would be resolved soon.

Bao

ADD REPLYlink written 12 months ago by Bao Ho80

Hi Jennifer,

Just tried and the problem persists.

Seems similar with the issues raised by Bao.

My workflow is NOT top secret :) and may be found here: https://usegalaxy.org/u/mike-2/w/diff1

Thanks for investigating!

Mike

ADD REPLYlink written 12 months ago by dragon2830

There seems to be a problem with input datasets being copied to a new history (when that is a selected option). This is causing the outputs of other steps (like FastQC) to be sent to the mapping tool, instead of the fastq data. Reviewing with team to see if a known issue.

A work-around could be to not send the results to a new history - testing that now.

More feedback soon & thanks for reporting the problem. Jen

ADD REPLYlink written 11 months ago by Jennifer Hillman Jackson23k

Running the workfow in the same history as the inputs does use the correct inputs, but the launched jobs are not complete. There are several workflow changes in progress and this may be related. No ticket yet, team reviewing, but I will post one back (or let you know that this is already fixed) once more is known. Jen

ADD REPLYlink written 11 months ago by Jennifer Hillman Jackson23k
1
gravatar for dragon28
11 months ago by
dragon2830
dragon2830 wrote:

Hi All, Just re-ran and the problem persists (nothing changed; still unable to run workflow). I waited some time for things to settle after the system changes. Any idea to to overcome? Mike

p.s. oddly, my prior message about the workflow disappeared. here is the workflow: https://usegalaxy.org/u/mike-2/w/diff1

ADD COMMENTlink written 11 months ago by dragon2830

I'll take a look at this Boa and Mike and get back to you both here. Is there a share history link you can share that contains the workflow inputs? Also, a shared link to a history that contains the problematic execution would be helpful. Both can be the same - as every history executed with a workflow automatically copies the input datasets over. Thanks, Jen

ADD REPLYlink written 11 months ago by Jennifer Hillman Jackson23k
0
gravatar for dragon28
11 months ago by
dragon2830
dragon2830 wrote:

Hi Jennifer, I just have two pairs of files (2x LAA and 2x LV). That is all which is being used for the inputs. The error is coming about as a result of trying to run the workflow; no error messages are being shown in the history (doesn't get that far). workflow-->diff1-->run-->choose files-->execute --> unable :(

Sincerely, Mike

ADD COMMENTlink written 11 months ago by dragon2830

I found the inputs and made a copy to test the workflow. Thanks. Jen

ADD REPLYlink written 11 months ago by Jennifer Hillman Jackson23k

The workflow has a few issues:

  • Inputs are single-end fastq files? Only one is sent to Tophat and Cufflinks. And just that single Cufflinks output is sent to Cuffdiff. Cuffdiff also needs at least two input BAMs. Otherwise, there is no comparison between datasets.
  • No Tophat results are linked into the Cuffdiff tool. This is the root of the problem during workflow execution.
  • The Cuffdiff conditions are also not given names in the workflow. This will cause problems once the above is adjusted.
  • Adding in all Cufflinks results (all dataset) to Cuffmerge to create the combined reference GTF to Cuffdiff is the one correct usage. Or use a public reference GTF (if the discovery of novel splices identified in the inputs do not wish to be considered).

Help:

ADD REPLYlink written 11 months ago by Jennifer Hillman Jackson23k
0
gravatar for dragon28
11 months ago by
dragon2830
dragon2830 wrote:

Great! Thank! Hope you can help with the resources available.

ADD COMMENTlink written 11 months ago by dragon2830
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 117 users visited in the last hour