m6A modification prediction using Nanopore DRS data (RNA004)

Question

[baibhu1234]

Hi everyone, I was doing m6A site modification prediction using m6Anet in a plant species. m6Anet have two models for RNA002 chemistry for human and arabidopsis.

but recently they have updated the m6net even for RNA004 chemistry. but the model is trained only on human cell line data.

which model i should use for predicting m6A sites in my plant? should it be RNA004 model trained with human data or RNA002 model trained with arabidopsis data.

Answer 1

[GenoMax]

If you have the complete data folder you should consider using dorado provided by ONT to do the basecalling. dorado should pick the correct model applicable for your data as long as there is direct internet access.

[baibhu1234]

thankyou,

I was trying the dorado tool for the m6A modification detection , but was getting segmentation error

dorado basecaller rna004_130bps_hac@v5.1.0 /scratch/XXXX/baibhav/ONT_DRS_data/20240911_DR6_R2_RNASeq_ONT/pod5 --modified-bases-models rna004_130bps_hac@v5.1.0_m6A_DRACH@v1/ > PAY11012_4ea26581_d051b054_0.bam

[2024-12-02 19:09:54.362] [info] Running: "basecaller" "rna004_130bps_hac@v5.1.0" "/scratch/XXXX/baibhav/ONT_DRS_data/20240911_DR6_R2_RNASeq_ONT/pod5" "--modified-bases-models" "rna004_130bps_hac@v5.1.0_m6A_DRACH@v1/"
[2024-12-02 19:09:54.388] [info] > Creating basecall pipeline
[2024-12-02 19:09:54.972] [info]  - BAM format does not support `U`, so RNA output files will include `T` instead of `U` for all file types.
Segmentation fault (core dumped)

[GenoMax]

Looks like you are either running out of memory or have a GPU/CPU that is unsupported.

Can you try running

dorado basecaller -r hac,6mA /scratch/XXXX/baibhav/ONT_DRS_data/20240911_DR6_R2_RNASeq_ONT/pod5/ > PAY11012_4ea26581_d051b054_0.bam

[baibhu1234]

hi GenoMax i was able to run dorado finally time dorado basecaller rna004_130bps_hac\@v5.1.0 /pod5/ --modified-bases-models rna004_130bps_hac\@v5.1.0_m6A_DRACH\@v1/ --device cuda:all --reference ref.fa > sample1.bam

it generated bam file for each replicate i did convert the bam to bed using "modkit pileup" now i want to all the downstream analysis for m6A modification (transcript/genome site level analysis).... such as getting the most confident sites and site ratio for each replicate

Then at last i would like to get the DMR analysis between condition based on 3 replicates each.