philippesanio
commented
3 months ago Hi @sandratra-rab
Thanks for raising that question.
Is it normal to get this error for MT chromosome specifically ? I have read the issues https://github.com/fritzsedlazeck/Spectre/issues/2 and https://github.com/fritzsedlazeck/Spectre/issues/29 but MT chromosome is indeed in my reference fasta file with the correct name written for the --only-chr parameter and MT sequence does not seem to contain any N regions.
No, this should not be specific to MT per see.
According to your description, I am guessing that there is no chromosome MT in the sample coverage data. Could you be so kind and confirmation for me if there is a chromosome MT in Mosdepth coverage data? You can check that by either looking at the in the Mosdepth sample summary file or searching in the sample coverage file regions.bed.gz.
Also, could you be so kind and check which version of Spectre you are running? spectre version. If you want the latest development version, please clone the Github repo and install Spectre locally using pip.
When running the same commandline without the --only-chr option, I get no error but no PNG file for MT chromosome is generated neither (same for various alt chromosomes inside my reference fasta file by the way, only chr 1-22, X, Y and only one alternative chromosome got a PNG output and I don't understand how it got selected).
Spectre generates only a coverage plot for every chromosome that is available for analysis.
Just a side note, please make sure to use the blacklist as it masks problematic regions, especially around the telomeres and centromeres. It is based on the gap data from UCSC grch38_blacklist.bed. We recommend using the grch38_blacklist_spectre.bed has been modified by expanding the masked regions, since a lot of samples we have tested showed poor coverage signals in those regions. In your case, you might have to change the chromosome names in the blacklist from eg "chr1" to "1". Not using the blacklist might increase your FPs dramatically, especially in the difficult regions. If you need different gap data, feel free to modify them.
Also, if you want to run Spectre on different samples with the same reference, make sure to tell Spectre with --metadata where it can find the MDR file. Otherwise, Spectre will calculate the same MDR file for every sample you are using again and again, which will cost a lot of time.
Best, Philippe
sandratra-rab
Hi,
I was testing Spectre latest release using mosdepth output and by submitting the following commandline:
CNVCaller --coverage test_Q20.regions.bed.gz --sample-id test --output-dir CNV_Calling/spectre/test/ --reference Ensembl/GRCh38.109/homo_sapiens.GRCh38.109.fasta --only-chr 1,10,11,12,13,14,15,16,17,18,19,2,20,21,22,3,4,5,6,7,8,9,MT,X,YI knew it should output PNG pictures for every chromosomes requested with the
--only-chrparameter (even when no particular long CNV would be found). However when reaching "MT" chromosome turn, i got this error:KeyError: 'MT'While generating the MDR file from the Ensembl GRCh38 r109 fasta file, Spectre returned this specific line regarding the MT chromosome (it worked perfectly fine for the previous chromosome numbers):
Is it normal to get this error for MT chromosome specifically ? I have read the issues #2 and #29 but MT chromosome is indeed in my reference fasta file with the correct name written for the
--only-chrparameter and MT sequence does not seem to contain any N regions.When running the same commandline without the
--only-chroption, I get no error but no PNG file for MT chromosome is generated neither (same for various alt chromosomes inside my reference fasta file by the way, only chr 1-22, X, Y and only one alternative chromosome got a PNG output and I don't understand how it got selected).Thank you in advance for your help!