According to different levels of m⁶A confidence, we extracted the corresponding m⁶A-associated variants as follows:
1． For m⁶A sites with high confidence level, we retained the variants that located nearby the m⁶A sites and then looked for the variants that destroy m⁶A motif of the m⁶A sites.
2． For m⁶A sites with medium confidence level, the m⁶A-associated variants were derived from the intersection between the variants and m⁶A sites generated from MeRIP-Seq experiments. Then, we applied our prediction models to find the variants nearby the m⁶A sites that change the sequence features essential for m⁶A modification.
3． For m⁶A sites with low confidence level, a genome-wide prediction based on Random Forest algorithm was performed for the sequences around all the collected variants from dbSNP and TCGA to find the variants that cause potential gain or loss of m⁶A sites.

In m6AVar, you can search or browse:

1) m⁶A-associated genetic mutations (dbSNP)

2) m⁶A-associated cancer somatic mutations (TCGA)

3) Disease related m⁶A-associated variants (GWAS and ClinVar)

4) Splicing sites affected by m⁶A-associated variants

5) RNA binding protein affected by m⁶A-associated variants

6) miRNA targeting and processing affected by m⁶A-associated variants

We collected millions of variants from dbSNP and TCGA project, as well as thousands of m⁶A sites from 7 miCLIP experiments, 2 PA-m⁶A-Seq experiments and 244 MeRIP-Seq experiments, together with a large number of predicted m⁶A sites are involved. At present, m6AVar contains 414,241 m⁶A-associated variants, which are composed of 352,014 m⁶A-associated germline mutations from dbSNP and 62,227 m⁶A-associated somatic mutations from TCGA.