MEME analysis issue

[hchintalapudi]

Hi,
I'm having issues with the MEME analysis for de novo motif finding.
Could you help me with what the issue is?

Thanks for the amazing tool!
slurm-97220.out.txt

Originally posted by @hchintalapudi in #14 (comment)

[fl-yu]

Hi Thanks for your interest of our tool!
It seems that meme did not work properly in your case, the following is message of error first begins.

Log::Log4perl configuration looks suspicious: No loggers defined at /dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/lib/site_perl/5.26.2/Log/Log4perl/Config.pm line 325.
Starting getsize: getsize random1000/MEME_SL_04_S4_shuf/SL_04_S4_summits_padded.fa 1> $metrics
/dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/libexec/meme-5.0.2/getsize: error while loading shared libraries: libicui18n.so.58: cannot open shared object file: No such file or directory
getsize exited with error code 127
getsize failed me... at /dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/bin/meme-chip line 748.

Sometimes there will be confilcts between meme and other software, do you install meme using conda by our tutorial?

[hchintalapudi]

I followed the conda isntallation of all dependencies and when I got this error I checked the installed pacakges in my conda env using "conda list" and did not find "meme".
So I checked this issue #2 and installed meme using conda install.

And meme is installed in my cutruntools2 env.

(cutruntools2) [f0042dk@discovery7 CutNrun]$ meme
 Usage:	meme	<dataset> [optional arguments]

	<dataset> 		file containing sequences in FASTA format
	[-h]			print this message
	[-o <output dir>]	name of directory for output files
				will not replace existing directory
	[-oc <output dir>]	name of directory for output files
				will replace existing directory
	[-text]			output in text format (default is HTML)
	[-objfun classic|de|se|cd|ce]	objective function (default: classic)
	[-test mhg|mbn|mrs]	statistical test type (default: mhg)
	[-use_llr]		use LLR in search for starts in Classic mode
	[-neg <negdataset>]	file containing control sequences
	[-shuf <kmer>]		preserve frequencies of k-mers of size <kmer> 
				when shuffling (default: 2)
	[-hsfrac <hsfrac>]	fraction of primary sequences in holdout set 
				(default: 0.5)
	[-cefrac <cefrac>]	fraction sequence length for CE region 
				(default: 0.25)
	[-searchsize <ssize>]	maximum portion of primary dataset to use
				for motif search (in characters)
	[-maxsize <maxsize>]	maximum dataset size in characters
	[-norand]		do not randomize the order of the input 
				sequences with -searchsize
	[-csites <csites>]	maximum number of sites for EM in Classic mode
	[-seed <seed>]		random seed for shuffling and sampling
	[-dna]			sequences use DNA alphabet
	[-rna]			sequences use RNA alphabet
	[-protein]		sequences use protein alphabet
	[-alph <alph file>]	sequences use custom alphabet
	[-revcomp]		allow sites on + or - DNA strands
	[-pal]			force palindromes (requires -dna)
	[-mod oops|zoops|anr]	distribution of motifs
	[-nmotifs <nmotifs>]	maximum number of motifs to find
	[-evt <ev>]		stop if motif E-value greater than <evt>
	[-time <t>]		quit before <t> CPU seconds consumed
	[-nsites <sites>]	number of sites for each motif
	[-minsites <minsites>]	minimum number of sites for each motif
	[-maxsites <maxsites>]	maximum number of sites for each motif
	[-wnsites <wnsites>]	weight on expected number of sites
	[-w <w>]		motif width
	[-minw <minw>]		minimum motif width
	[-maxw <maxw>]		maximum motif width
	[-allw]			test starts of all widths from minw to maxw
	[-nomatrim]		do not adjust motif width using multiple
				alignment
	[-wg <wg>]		gap opening cost for multiple alignments
	[-ws <ws>]		gap extension cost for multiple alignments
	[-noendgaps]		do not count end gaps in multiple alignments
	[-bfile <bfile>]	name of background Markov model file
	[-markov_order <order>]	(maximum) order of Markov model to use or create
	[-psp <pspfile>]	name of positional priors file
	[-maxiter <maxiter>]	maximum EM iterations to run
	[-distance <distance>]	EM convergence criterion
	[-prior dirichlet|dmix|mega|megap|addone]
				type of prior to use
	[-b <b>]		strength of the prior
	[-plib <plib>]		name of Dirichlet prior file
	[-spfuzz <spfuzz>]	fuzziness of sequence to theta mapping
	[-spmap uni|pam]	starting point seq to theta mapping type
	[-cons <cons>]		consensus sequence to start EM from
	[-brief <n>]		omit sites and sequence tables in
				output if more than <n> primary sequences
	[-nostatus]		do not print progress reports to terminal
	[-p <np>]		use parallel version with <np> processors
	[-sf <sf>]		print <sf> as name of sequence file
	[-V]			verbose mode
	[-version]		display the version number and exit

[fl-yu]

Yes that is the point! So I recommend you to create another environment for meme installation as shown in the tutorial. Hope this helps.

[hchintalapudi]

I actually followed these steps using environment1.yml and environment2.yml in the install dir
I have two separate envs like you mentioned above.
Yet I get these errors.
slurm-105153.out.txt

That's when I decided to conda install meme again in the cutruntools2 env (that was ignorant I guess)
I deleted my existing cutruntools2 env from environment1.yml and installing it all over again.

I'll post if the issue still persists!
Thanks.

[fl-yu]

Can you tell me why the message begins with

Could not find conda environment: cutruntools2
You can list all discoverable environments with `conda info --envs`.

The error is same, and I believe it was becasue of the conficts between meme and other libraries. it seems that the meme you specified which was installed with other software in the same environment. Make sure the path of software sepcificed in the configuration file are the ones you wanted. Also please post your configuration file?

[hchintalapudi]

So, I was trying to activate the env in my slurm scheduler script:

#!/bin/bash
###################################
# Name of the job
#SBATCH --job-name=CutNrun
###################################
# Number of compute nodes
#SBATCH --nodes=1
###################################
# Number of cores
#SBATCH --ntasks-per-node=40
###################################
# Walltime (job duration)
#SBATCH --time=10:00:00
###################################
# Account to use
#SBATCH --account=nccc
###################################
# Name of queue / partition
#SBATCH --partition=standard
###################################
# Email address
#SBATCH --mail-user=himanshuchintalapudi@gmail.com
###################################
# Email notifications
#SBATCH --mail-type=BEGIN,END,FAIL

source /optnfs/common/miniconda3/etc/profile.d/conda.sh
conda activate cutruntools2

And I get that error message:

Could not find conda environment: cutruntools2
You can list all discoverable environments with `conda info --envs`

However I commented it out and the pipeline runs fine, no issues! (at least till the MEME part fails)

My config file attached.
bulk-config.json.txt

[fl-yu]

There is no problem with the configuration.
Can you test if getsize function from meme works? Test this before and after activate meme environment.
/dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/libexec/meme-5.0.2/getsize

[hchintalapudi]

[f0042dk@discovery7 CutNrun]$ /dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/libexec/meme-5.0.2/getsize
/dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/libexec/meme-5.0.2/getsize: error while loading shared libraries: libicui18n.so.58: cannot open shared object file: No such file or directory
[f0042dk@discovery7 CutNrun]$ conda activate meme
(meme) [f0042dk@discovery7 CutNrun]$ /dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/libexec/meme-5.0.2/getsize
/dartfs-hpc/rc/home/k/f0042dk/miniconda3/envs/meme/libexec/meme-5.0.2/getsize: error while loading shared libraries: libicui18n.so.58: cannot open shared object file: No such file or directory

:/

[fl-yu]

It seems that the conflicts still exist even you installed the software separately. I am sorry can not help this further, but you may ask for help from your IT admin to make sure meme works properly.

[ShikanZ]

I had the same issue.
(meme) shikan@shikan-p520:/mnt/Data/Cut_and_tag_nf/docker$ getsize
getsize: error while loading shared libraries: libicui18n.so.58: cannot open shared object file: No such file or directory

I figured out installing this missing library using conda. First, remove your current meme environment. Then install using this command, conda create -n meme python=3.6 icu=58 meme

Hope it helps!

Shikan

[hchintalapudi]

I believe I tried this after looking at some old issues here.
But will work on this again and post an update, thank you @ShikanZ.

[fl-yu]

Hi Guys, for meme conflicts, can you try this solution, which I have tested in a clean environment on the cluster.

conda create -n meme meme=5.0.5 python=3.7
conda activate meme

If you did not set the conda channel yet, run these commands first

conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge

Please let me know if this works for you

[hchintalapudi]

Hi @fl-yu,
I tried what you suggested above, the other week and still got some errors.
slurm-138333.out.txt