Trouble Running Snakemake via Tibanna #256
Comments
|
The log file says Do you mind running with |
|
So far on the screen: [i also do not have a docker for this, I am assuming it will use software installed from ec2 that I kicked off snakemake] job input tibannatestdata/case6_WEX_T_sample.sorted.bqsr/case6_WEX_T_sample.sorted.bqsr.bam |
|
Please let me know if you want snakefile as well. Thanks! |
|
Yes that would be helpful! |
|
Do you use conda for installing dependencies? (If so, you'd need to add |
|
Attached I am manually installing dependencies, am not using conda. In general, if software dependencies are installed correctly in instance that I am lauching tibanna, do I need a docker or conda environment? Thanks for your help! |
|
snakemake version is 5.7.4 |
|
I see. In terms of dependencies, you'd have to then use |
|
In terms of log file not going to the S3 bucket, the directory must be under the current working directory ( |
|
I'm not totally sure why snakemake could not find the rule. I will refer to this to snakemake in case Johannes has a better idea. |
|
Ah, you can run |
|
Would I be able to point the per-command to a bash script because there are
several programs that need to be installed. Regarding snakemake specific
issues I will troubleshoot more and ask Johannnes. It is ok if we keep this
issue open for now? Thank you for taking the time to help me out.
…On Thu, Nov 14, 2019 at 5:57 PM Soo Lee ***@***.***> wrote:
I'm not totally sure why snakemake could not find the rule. I will refer
to this to snakemake in case Johannes has a better idea.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#256?email_source=notifications&email_token=AI254Z2QL3FUEVZ5ON7IQU3QTXJVVA5CNFSM4JNSD45KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEDTXQA#issuecomment-554122176>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AI254Z5THCSQQBJJQ7Y3GBLQTXJVVANCNFSM4JNSD45A>
.
|
|
If the directory is a git repo, the shell script will also be automatically passed to an EC2 instance along with the other files, so it would be safe to call the shell script in |
|
Sure, we can keep this issue open until it's resolved. |
|
As the error says, check first if your input function of the rule read_counter does not produce an exception. Just convert the lambda expression into a real function and debug into it or print any raised exceptions. Maybe that sample identifier cannot be found in your config file, thus raising a key error? |
|
Hi Johannes, The dry run worked: rule read_counter: rule read_counter: However it produces this same error: raise Exception("output file {} upload to {} failed. %s".format(source, bucket + '/' + target) % e) I am still getting same error as before. Do you know why it adds "/data1/snakemake" to the input path? I do no not have that path in the samples.yaml. Thank you ! @SooLee Thank you! I will continue to troubleshoot and post comments here. Thank you for your help |
|
tibannatest is name of s3 bucket |
|
@sengoku93 |
|
I have attached new Snakemake and corresponding samples.yaml. I am not sure how to resolve this error. I have changed lambda function and am still gettin same error. Thanks |
I am currently using snakelike --tibanna on an EC2. Is that okay? |
|
I have attached updated snakemake where I am explicitly stating remote paths. However, I am still getting same error as earlier. Please let me know if you have any suggestions. Apologies for multiple comments. Thank you! |
|
Last quick question, can you load a specific AMI when tibanna launches instances to execute snakemake? Thanks |
|
Hi @SooLee , I was able to get log file successfully copied to S3. The errors I get from the log file is this: /bin/bash: readCounter: command not found Which means the Conda environment is not copying over to correct instance. I then updated snakemake to explicitly have Conda with corresponding environment.yaml. I am still getting same above error. Regarding snakemake error that started this whole post (#256), I believe it is due to this error I am getting now. Command line for updated snakemake: snakemake --tibanna --default-remote-prefix=tibannatestdata -np -s tibanna_ichorCNA_snakemake.mk --use-conda v2_tibanna_ichorCNA_snakemake.txt Thank you! |
|
Hi, |
|
@sengoku93 Sorry I didn't get notified about the updates of this issue. I will take a look today. |
|
I resolved issue. Thank you! |
|
I do have a question regarding how to specify Snakemake to look through folder of choice. For example if I want to run bwa mem, I would need reference hg38.fa which can be a param in snakemake. However, bwa mem needs the hg38.fa.fai file as well but since it is not specified in the snakemake, Tibanna can't find it. Would I put the .fai file as an input or a param to make sure tibanna can look for it appropriately? Thanks |
|
For example can I have a config.json as you would with tibanna+cwl, but instead for tibanna+snakemake? I do see that snakemake said it will support --configfile argument but I don't see any information if it is currently supported. Thank you! |
|
When running this command line: snakemake --tibanna --default-remote-prefix=tibannatestdata -np -s v1_alignment.mk --use-conda --default-resources "mem_mb=max(4input.size, 10000)" "disk_mb=max(4input.size, 10000)" I get the following error: [bwt_restore_bwt] Failed to allocate 3209286160 bytes at bwt.c line 452: Cannot allocate memory I believe it is because --default resources is not working as it should. I am using a test fastq with 30mb r1 and 30mb r2 fastqs. Is there a way to specify instant to be used? Or documentation regarding allocating memory? Thank you! |
|
@sengoku93 |
|
I will close this so that I can get notified when you reopen it. Sorry for the inconvenience. |
|
When running with --verbose option: Building DAG of jobs... [Tue Dec 10 15:56:02 2019] job input tibannatestdata/case6_WEX_T_sample/fastq/case6_WEX_T_sample.read1.fastq [Tue Dec 10 15:56:02 2019] job input tibannatestdata/case6_WEX_T_sample/fastq/case6_WEX_T_sample.read1.fastq I will try with config file. Is there documentation describing config file specifically for snakemake? Thanks |
|
I can not reopen issue after you closed it. |
|
It looks like you requested 22Tb of memory. Was that intended? How large is your input file? |
|
When using configfile where I specified the type of instance used ( in this case c2.5xlarge): snakemake --tibanna --default-remote-prefix=tibannatestdata -s v1_alignment.mk --use-conda --configfile=config.json I get the error saying config file not found which makes sense because tibanna launches t3.micro instead of c2.5xlarge. |
|
My input files are two truncated fastqs that are 30mb each. |
|
Tibanna doesn't get the instance type from config. It will auto-determine the instance type based on cpu and memory requirements. The memory requirement calculation seemed wrong base on what I see. I opened an issue in the snakemake repo. (Memory calculation is global for snakemake and not specific for tibanna.) Can you try specifying the memory requirement inside the config file? |
|
Ah and the config file must match the format of snakemake. |
|
Is there an example for the config file format for snakemake? |
|
Also, even though the fastq files were 30mb. I also had reference genome files that totaled to about 4gb. I am not sure if those are incorporated into the memory calculation. Thanks |
|
Actually yes, all the input files will be considered in the memory calculation. |
|
To clarify can you describe what the command below will do? (i.e what does the 10000 input do) |
|
https://github.com/NIH-HPC/snakemake-class/blob/master/exercise06/Snakefile This is a random example snakefile and you can see how they specified memory inside this file. A config file can be used to redefine variables in snakefile. It should be somewhere in the Snakemake documentation. |
|
The option means it will choose maximum between 4 x input size and 10,000mb. (and same for disk space) |
|
Okay thanks, I will look at the link you sent. So currently, there is not a way to specify instance type we would like to use? Thanks |
|
nope, not through snakemake. |
|
Okay, no problem. Would tibanna + cwl offer option to choose instance type? |
|
what does the log say? Yes with cwl, you can specify the instance types. |
Dear sengoku93, would you remember how you fixed the issue ? I am currently having the same one and would like to understand where I am going wrong. I think I am also setting my conda incorrectly. If you have any information in that regard, it would be greatly appreciated. Thanks ! :) |
Hi,
I am running custom snakemake via tibanna. However, the runs fail and it does not deposit snakemake log to corresponding s3 bucket I deployed. I have attached log file here for your review. Any suggestions.
This is the command I used: snakemake --tibanna --default-remote-prefix=tibannatestdata -s tibanna_ichorCNA_snakemake.mk
Thank you!
yuWQY5gCrdfz.log
The text was updated successfully, but these errors were encountered: