Hi everyone,
first time posting an issue. First thank you for having developed and keeping updating modkit, extremely useful for my research. Indeed, I am interested in RNA m5c methylation. I have ONT generated data from 2 cell lines, namely a control condition and a line knockdown (KD) for NSUN2, a major m5c methylator. In particular I added the latter as negative and quality control, to be sure that the called m5c sites disappeared upon NSUN2 KD. I have used the following lined of codes for modkit pileup
$modkit pileup
-t "${SLURM_CPUS_PER_TASK}"
--filter-threshold 0.8
--mod-threshold C:0.99
--modified-bases C:m
--reference "$REF" \ # alignment to genome
"${input_bam}"
"${output_dir}/${sample_name}.raw.bed"
--log-filepath "${output_dir}/${sample_name}.pileup.log"
Then I use an inhouse R script for differential methylation analysis as well as modkit dmr pair with the following commands
$modkit dmr pair
-a "$STG/Pilot_sample1/Pilot_sample1.raw.bed.gz" . # a CTRL condition
-a "$STG/Pilot_sample2/Pilot_sample2.raw.bed.gz"
-b "$STG/Pilot_sample5/Pilot_sample5.raw.bed.gz" \ # B NSUN2 KD
-b "$STG/Pilot_sample6/Pilot_sample6.raw.bed.gz"
--ref "$REF"
--base C
--single-code m
--cap-coverages
--min-valid-coverage 20
--header
-o "$DMR/control_vs_siNSUN2.dmr.bed"
--log-filepath "$DMR/control_vs_siNSUN2.dmr.log"
-t "${SLURM_CPUS_PER_TASK}"
I have been trying with multiple combination of thresholds but I keep getting thos kinds of results (the plot shows the significantly differentially methylated m5c sites between NSUN2 KD vs CTRL plotted according to the stoichiometry in the CTRL condition). From the biological perspective, It unlikely that I have more site in the NSUN2 KD condition, so I don t know what could have gone wrong (biologically I am sure of the knockdown of the protein, checked by different other analysis).
Could you help me? Thanks
Marcello
Hi everyone,
first time posting an issue. First thank you for having developed and keeping updating modkit, extremely useful for my research. Indeed, I am interested in RNA m5c methylation. I have ONT generated data from 2 cell lines, namely a control condition and a line knockdown (KD) for NSUN2, a major m5c methylator. In particular I added the latter as negative and quality control, to be sure that the called m5c sites disappeared upon NSUN2 KD. I have used the following lined of codes for modkit pileup
$modkit pileup
-t "${SLURM_CPUS_PER_TASK}"
--filter-threshold 0.8
--mod-threshold C:0.99
--modified-bases C:m
--reference "$REF" \ # alignment to genome
"${input_bam}"
"${output_dir}/${sample_name}.raw.bed"
--log-filepath "${output_dir}/${sample_name}.pileup.log"
Then I use an inhouse R script for differential methylation analysis as well as modkit dmr pair with the following commands
$modkit dmr pair
-a "$STG/Pilot_sample1/Pilot_sample1.raw.bed.gz" . # a CTRL condition
-a "$STG/Pilot_sample2/Pilot_sample2.raw.bed.gz"
-b "$STG/Pilot_sample5/Pilot_sample5.raw.bed.gz" \ # B NSUN2 KD
-b "$STG/Pilot_sample6/Pilot_sample6.raw.bed.gz"
--ref "$REF"
--base C
--single-code m
--cap-coverages
--min-valid-coverage 20
--header
-o "$DMR/control_vs_siNSUN2.dmr.bed"
--log-filepath "$DMR/control_vs_siNSUN2.dmr.log"
-t "${SLURM_CPUS_PER_TASK}"
I have been trying with multiple combination of thresholds but I keep getting thos kinds of results (the plot shows the significantly differentially methylated m5c sites between NSUN2 KD vs CTRL plotted according to the stoichiometry in the CTRL condition). From the biological perspective, It unlikely that I have more site in the NSUN2 KD condition, so I don t know what could have gone wrong (biologically I am sure of the knockdown of the protein, checked by different other analysis).
Could you help me? Thanks
Marcello