Generate 3D images of flu HA proteins with variants highlighted

[Deena-B]

Background

3D protein structures hold richer information than nucleotide sequences. Images can be analyzed using CNNs. Need I say more? Let's make some images!

Aim

Generate about 5 panels of 3D images of flu H1 HA proteins with amino acid variants highlighted

Method

1. Find a good PDB structure for influenza H1N1 A/California/4/2009 or A/California/7/2009. Good is defined by:

• not having an antibody bound
• no (or few) missing residues when the surface view of the protein trimer is spun 360 degrees

2. Download human H1N1 HA sequences from the 2013-14 flu season (see fluHA issue 1 on where to get the sequences)
3. Align & cluster using a hierarchical tree method
4. Create consensus sequences for the whole season and a consensus sequence for each major cluster
5. Use Pymol to generate 3D images using each cluster's consensus with mutated residues from where they differ from the PDB structure and color the residues that vary from the consensus of the whole season
6. Take X number of snapshots of each HA from different angles and degrees of rotation and put the snapsots into a single 'panel' image

Document your work

Please fork & clone this repo. Check out a branch with the issue number (e.g. 3Di2) for your work, then push and make a PR for us to merge.

Add a note with your process to the nb folder (nb stands for notebook). Your note can be .md or .ipynb or other types. For an example directory structure, see overview/wiki dir-struct.

Your note should be named as follows:

• DATE-issue#-shortdescription.ext

e.g.:

• 20190701-i02-extract-chrM-fa.md

If you keep large files within the fluHA repo, remember to add their extension to the .gitignore file. If you can make an example file that is small, then use the annotation .eg.ext and create a .gitignore exemption with !. (See mitolin .gitignore for examples)

Questions?

Please put questions related to this issue in this issue thread. If you want a quick response, post a link to your comment in this thread to Slack #deepcelllineage or DM @deena. To join Slack enter your email address here. For questions NOT specifically related to this issue, get in touch through any of the communication methods listed in DCL's overview README.