Unperfomed analyses despite no warning message in TORMES log file?

[m-ocejo]

Dear Narciso, as I previously mentioned in past issue #11 , we encountered several errors after running TORMES pipeline. We didn´t obtain any output for Pangenome and FimH analyses. Furthermore, species identification with Kraken gave unclassified results. We were wondering about the possible cause, as the tormes log did not display any warning message. Please find attached the log file: tormes.log

Thank you!!

[jllavin77]

Dear Narciso Thnak you for creating this wonderful pipeline and take the time to prepare the conda package. I'm facing the same issue myself and would like to ask you a couple of questions. Is there anything required that we should install apart from the programs in the conda container? Otherwise I cannot understand why some parts of the pipeline like pangenome or FimH analyses are skipped...

Best regards

JL

PD. I also faced the Mauve issue but apparently was able to solve it after stating the correct version of the program in the config file...

[m-ocejo]

Dear Narciso, I tried to run the Kraken separately to see if the error might be in the installation. It seems that TORMES skipped the kraken assignation due to the lack of options in the piped kraken_report command. In line 795 of the TORMES code (https://github.com/nmquijada/tormes/blob/master/bin/tormes):

$KRAKEN --threads $CPUS --db $KRAKENDB --preload --only-classified-output --paired $OUTWD/cleaned_reads/$i\.ok_1.fastq.gz $OUTWD/cleaned_reads/$i\.ok_2.fastq.gz | $KRAKENREPORT > $OUTWD/species_identification/$i\_identification.txt

when I add the option --db $KRAKENDB it worked perfectly.

$KRAKEN --threads $CPUS --db $KRAKENDB --preload --only-classified-output --paired $OUTWD/cleaned_reads/$i.ok_1.fastq.gz $OUTWD/cleaned_reads/$i.ok_2.fastq.gz | $KRAKENREPORT --db $KRAKENDB > $OUTWD/species_identification/$i_identification.txt

Also, as mentioned in the previous message of this thread, I encountered Mauve.jar issue but also solved it after editing the config_file.txt as described in this solution https://github.com/nmquijada/tormes/issues/1#issuecomment-537957065.

Hope this might help. Thank you for the hard work!

--

[nmquijada]

Hi all, Indeed, the problem with Mauve is that conda installs the newer Mauve version (2.4.0.r4736-1) instead the one I used for creating the conda yaml file (2.4.0.r4736-0). The problem can be solved by changing the config file of tormes, but I included the solution for the automatic installation of next tormes version.

@m-ocejo thank you very much for sharing your solution for kraken. It is the first time I encounter someone with this issue, so I will implement your solution in the new version :)

I have also included more stringent options for the tormes.log file, as I found out that some errors are not reported.

I will work on why FimH typing and pangenome didn't work for you. I will need an "error log file" in order to inspect the errors that appear in you analysis. Could you please run the following command (by changing the name of the files and number of threads according to your analysis) and then share the error-tormes.txt file?

tormes --metadata YOUR-METADATA.txt --output YOUR-OUTPUT --threads 16 &>>error-tormes.txt 2>>error-tormes.txt &

Thanks!