Solving structure of Bacteriophage

Data Processing Strategy
1

Dear Cryo-EM Community,

I’m working on a cryo-EM dataset of a bacteriophage with a tail, and I’m encountering challenges due to the large particle size (>1200 Å, including the tail). I would greatly appreciate your guidance on the following:

  1. What is the best approach to pick such large particles, given their size and complexity?
  2. When using blob-based picking, how should I configure the parameters to optimize for these dimensions?
  3. Should I consider boxing the head and tail separately for better results, or are there alternative strategies that work well for such cases?

Your suggestions and tips based on your experience would be invaluable.

Thank you in advance for your insights!

Best regards,

Subhomoi

2

Depends on how many micrographs you have. <5000, I’d suggest doing it manually if the number of particles per micrograph is low (<10). It won’t take that long, and you can make sure you don’t have any crazy picks that 2D classification can’t seem to sort out (2D classification can struggle to differentiate between a 2800 Å virus and a 2800 Å ice blob…)

Yes. :slight_smile: That way you can use optimal symmetry for the head and tail independently (and symmetry expand a focussed refinement on the joint). Also you’ll keep box sizes more reasonable. Remember, CryoSPARC has a functional limit of 1120 pixel boxes, regardless of GPU memory.

3

Training a general model like Yolov11 will help you to identify micrographs with bacteriophages on them. Then use lower confidence threshold to box individual bacteriophages within micrographs. Perhaps you could use it to label subparticles you are after. You will need to manually label ~2000 features for good performance. You could start with a grid pick within each box (?).

Good luck!

4

Thank you for your suggestions !!