Hi, thanks for developing this great tool!
I’m working with Fiber-seq data in maize (Zea mays) and have a question regarding FIRE calling and model applicability across species.
As far as I understand, the current FIRE calling framework is largely developed and trained based on mammalian systems, where CTCF plays a central role in chromatin organization. However, in plants like maize, CTCF is absent, and chromatin architecture is known to be quite different.
This raises a couple of questions for me:
1. Is it necessary (or strongly recommended) to train a species-specific model for maize when performing FIRE calling with Fiber-seq data?
2. If so, what kind of training data would be required to build a reliable maize-specific model?
For example:
• Do we need labeled FIRE regions derived from orthogonal assays (e.g., Hi-C, ATAC-seq, MNase-seq)?
• Is high-coverage Fiber-seq data alone sufficient for training?
• Are there any guidelines on minimum data scale or feature requirements?
3. More generally, how dependent is the current FIRE model on CTCF-related features, and how well would it be expected to generalize to plant genomes?
Any suggestions or best practices for applying FIRE calling in non-mammalian systems (especially plants) would be greatly appreciated.
Thanks a lot for your help!
Hi, thanks for developing this great tool!
I’m working with Fiber-seq data in maize (Zea mays) and have a question regarding FIRE calling and model applicability across species.
As far as I understand, the current FIRE calling framework is largely developed and trained based on mammalian systems, where CTCF plays a central role in chromatin organization. However, in plants like maize, CTCF is absent, and chromatin architecture is known to be quite different.
This raises a couple of questions for me:
1. Is it necessary (or strongly recommended) to train a species-specific model for maize when performing FIRE calling with Fiber-seq data?
2. If so, what kind of training data would be required to build a reliable maize-specific model?
For example:
• Do we need labeled FIRE regions derived from orthogonal assays (e.g., Hi-C, ATAC-seq, MNase-seq)?
• Is high-coverage Fiber-seq data alone sufficient for training?
• Are there any guidelines on minimum data scale or feature requirements?
3. More generally, how dependent is the current FIRE model on CTCF-related features, and how well would it be expected to generalize to plant genomes?
Any suggestions or best practices for applying FIRE calling in non-mammalian systems (especially plants) would be greatly appreciated.
Thanks a lot for your help!