Killifish Hypoxia RRBS Part 19
Wrapping up the initial methylation analysis
I have a few loose ends to tie up before finishing my preliminary methylation analysis: methylation landscape information and an overview of the closest annotated genes.
Methylation landscape analysis
I tried looking through the output to figure out if this information was listed in a file anywhere. I didn’t have the log files anymore, but I don’t think it was listed there. In any case, if someone else isn’t able to access the information I used, then it wasn’t reproducible! So I returned to this Jupyter notebook to revise the methylation landscape analysis. Neel said that I should recalculate my statistics using 0% for the cutoff for methylation (i.e. a CpG was either methylated or unmethylated, and it didn’t matter how intensely it was methylated).
**Table 1. Methylation landscape information calculated using all common CpGs for each specific contrast.
Contrast | Methylated CpGs (%) | Unmethylated CpGs (%) | Average Methylation |
---|---|---|---|
All samples | 7275 (48.8%) | 7620 (51.2%) | 20.7% |
All NBH | 12216 (82.0%) | 2679 (18.0%) | 20.6% |
All SC | 12536 (84.2%) | 2359 (15.8%) | 20.7% |
Hypoxic NBH | 96857 (54.9%) | 79429 (45.1%) | 22.1% |
Normoxic NBH | 93895 (53.3%) | 82391 (46.7%) | 20.4% |
Hypoxic SC | 22694 (27.6%) | 59611 (72.4%) | 16.4% |
Normoxic SC | 45661 (55.5%) | 36644 (44.5%) | 16.2% |
Compared to what I have seen with other fishes, general methylation seems to be low. Interestingly, the tolerant NBH population doesn’t have much of a difference between the methylated:unmethylated ratios after hypoxia exposure, but the sensitive SC population does (~70% unmethylated in hypoxia vs. ~45% in normoxia). Average percent methylation is also lower in the SC population. Seeing population differences makes me think more about identifying SNPs between these populations and trying to identify mQTLs.
Closest genes to DMR
The next thing I wanted to do was dig into the functions of the closest genes to DMR. I used the ENSEMBL and NCBI databases to find this informaiton.
Overlapping genes:
- iqgap1 (IQ motif containing GTPase activating protein 1): enable GTPase activator activity, actin filament binding activity, and calmodulin binding activity, regulate cytoskeletal organization
- aspa (aspartoacylase): enable hydrolase activity
- SHISA6 (protein shisa-6 homolog): enable ionotropic glutamate receptor activity, part of AMPA glutamate receptor complex, involved in synaptic density and membrane
Closest genes:
- ENSFHEG00000011518 (kcnma1a; -22,011 bp away): enable calcium-activated potassium channel activity, acts within auditory stiumlus response, active in postsynaptic membrane
- ENSFHEG00000021357 (no gene name; -24,519 bp away): major histocompatibility complex class I-related gene protein-like (not predicted in the latest F. heteroclitus annotation)
- znf423 (-55,168 bp away): zinc finger protein 423, DNA-binding transcription factor, roles in signal transduction
The fact that one of the DMR was close to a gene not present in the latest mummichog annotation supports the idea of re-doing RRBS and RNA-Seq alignment with the new genome! Something to consider after doing an initial pass of the data.
Going forward
- Revise methylation landscape information
- Update methods and results
- Match DMR with RNA-Seq information
- Start mapping with new genome
- Try DMR identification with
bismark
andmethylKit
- Create OSF repository for all intermediate files