Jasmine

Call select base modifications in PacBio HiFi reads

---

jasmine identifies DNA base modifications in PacBio HiFi reads by analyzing polymerase kinetic signatures. The current models support detecting 5-Methylcytosine (5mC) at CpG sites, N6-Methyladenine (6mA) with strand-specific calls compatible with Fiber-seq tools such as fibertools (Jha 2024), and 5-Hydroxymethylcytosine (5hmC) with strand-specific calls at CpG sites. For 5mC, the caller assumes that methylation is symmetric at the CpG site and reports methylation on the forward strand. Results are written as standard MM/ML tags in the output BAM. All three callers are enabled by default.

Availability

Latest version can be installed via bioconda package pbjasmine.

Please refer to our official pbbioconda page for information on Installation, Support, License, Copyright, and Disclaimer.

Latest Version

Version 26.1.2: Full changelog here

Input Data

Input for jasmine is a PacBio HiFi BAM containing kinetics tags (fi/fp/ri/rp or ip/pw). For more info see ccs.how.

Reads with fewer passes than --min-passes (default 2) are written to the output without modification tags.

Execution

jasmine

Running jasmine is as simple as:

jasmine movie.hifi_reads.bam movie.methylation.hifi_reads.bam

pbmod

Utility multi-tool for inspecting, reporting, and comparing methylation calls.

# Per-base methylation table with ANSI color
pbmod inspect --color movie.jasmine.bam

# Compare modification calls between two BAMs
pbmod compare reference.jasmine.bam query.jasmine.bam

Output Data

The output methylation prediction for each annotated HiFi read is encoded in the MM and ML tags, defined in the SAM tag specification. The MM tag specifies the modification and to which base it applies. The ML tag specifies the probability of methylation at each base.

The output is also described in the PacBio BAM file format documentation as

Tag	Type	Description
MM	Z	Base modifications / methylation
ML	B,C	Base modification probabilities

Notes for ML: The continuous probability range of 0.0 to 1.0 is remapped to the discrete integers 0 to 255 inclusively. The probability range corresponding to an integer N is N/256 to (N + 1)/256. The ML tag presents the probabilities in the order of modifications seen in the MM tag.

The absence of MM/ML tags on a read indicates either no kinetics data (np = 0) or insufficient passes (below --min-passes). An empty MM/ML tag (present with no entries) means the read was processed and no modification sites were detected.

Example

Read  AGTCTAGACTCCGTAATTACTCGCCTAG...
C        1    2 34       5 6 78
CpG              *         *

MM:Z:C+m,3,1,...   # CpG sites are at C #4 (1+3) and #6 (1+3+1+1)
ML:B:C,249,4,...   # probability of methylation at the first CpG is in [249/256,250/256); second CpG is in [4/256,5/256).

Model Training

The jasmine methylation models are trained using a supervised learning approach with curated positive and negative datasets.

Model	Positive datasets	Negative datasets
5mCpG	HG002 WGA + M.SssI	HG002 WGA
6mA	HG002 Fiber-seq	HG002 WGS, HG002 Fiber-seq

For the HG002 Fiber-seq training dataset, positive and negative labels were generated using fibertools predict-m6A (Jha 2024).

Changelog

PacBio recommends the latest version for all Sequel II, Revio, and Vega datasets. PacBio recommends against using older versions.

• 26.1.2

  • New 5-Hydroxymethylcytosine (5hmC) caller at CpG sites (strand-specific calls)
  • New models for SPRQ-Nx chemistry (R/P2-C3/5.0-25M)
  • 6mA calling enabled by default
  • New pbmod multi-tool for inspecting, and comparing methylation calls

• 2.7.99

  • Fix by-strand 6mA calling

• 2.4.0

  • Updated 5mCpG caller, with new model design and input features
  • New 6mA caller

• 2.0.0

  • Initial release of jasmine 5mCpG model that supports Sequel II and Revio.
  • Support for single-strand consensus reads