Meeting notes

Having discussed the paper Critical Assessment of Metagenome Interpretation – a benchmark of metagenomics software, we looked over LINbase’s potential in metagenomics analysis.

Meeting notes

• Chengjie talked about the issues and things to improve he’s found in the website (see here).
• Long will finish adapting the website to the new database schema next week, and work on converting Taxonomy ID to LINgroups after that, hopefully done in another 2 weeks.
• Boris also has some new thoughts about the extension of identification functionality.

To-do list

Maintainability Issues

1. Change project name
2. Add .gitignore to exclude unnecessary files
3. Standardize file names
4. Remove uncessary files
5. Use .htaccess to restrict folder access
6. Change all absolute path to relative ones

LIN meeting updates

Research memo

• Mistakes in the model construction using Random Forest
  • Whether there is Jaccard similarity should be the first step
    • No way to put weight on a feature, can only split the data manually
    • The threshold that all ANI has corresponding jaccard similarity is 0.9, but we can still make it 0.95, to be conservative

Research memo

• Confirmed: There is no case with ANI<0.3 and cov<0.7, which confirms how ANI is calculated: only considers those alignments whose cov>0.7 & identity>0.3
• But tons of cases whose ANI’s pretty high but cov<0.7.
• When only looking at ANI pairs whose cov>70%, the linear correlation between ANI and estimated ANI is much better, 0.994.
• Among 280 pairs which are filtered out but having > 95% ANI, only 3 pairs not in the same species. Even more, 2 of the pairs belong to genera Escherichia and Shigella, and 1 pair belongs to Agrobacterium genus.
• Confirmed, as long as it’s > ANI 95%, however coverage is, belongs to the same species.
• What makes us confuse is the part that we cannot assign a 95% level LINgroup using MinHash because there’s no Jaccard similarity reported.
• Check tax ID from metadata, index=4, maybe use data mining to dig the taxonomy out.

Things to do after meeting

last update: 9/5/2017

• For adding LINgroup descriptions, tax- or non-taxonomic group is first class, and comments, urls are associated to that class (always show up). There might be multiple comments and urls for the same LINgroup, make it possible for display.
• Add “All” in LINgroup search
• Go to fix search pages if nothing can be found.
• Update LINgroup after LIN’s assigned
• Make profile page more powerful, so he/she can see what’s been submitted, genomes or LINgroups, and, edit them, and set it private or public.
• Add email link to LINgruop profile page at who described it.
• Relocate BLAST database to /home directory

What I’ve done today

• Karan changed LIN’s font and color for searchprocessview page, I applied the same function to all pages as long as LIN is displayed.
• Made a hierarchy look of LINgroup buttons
• Adjusted footer position, but there’s more to adjust…

Copied from the previous to-do list

Copied from the previous to-do list

Copied from the previous to-do list

To-dos today

• Boris found errors in LINbase:
  • LINgroup description only describes 0, which is the first digit.
  • Shows multiple descriptions to the same single LINgroup, which we don’t need.
  • Align LINgroups by hierarchy
  • Add nav bar to every page.
• Search by “All” is not up yet.
• Also I noticed that search by LINgroup description isn’t functioning properly.
• Avail “keyword” searching in SQL.
• Add “Family” to Description_Items Done today
• Change “Nickname” to “Non-taxonomic group” Done today
• While LINgroup description, user has to describe at least one of “family”, “genus”, “species”, “Non-taxonomic group” before getting to comment or add url.

Copied from previous dev update

Problems found

• GenomeProfile was using out-dated database. solved on 5/23/17
• The way UserProfile page lists submitted genomes is not friendly at all.
• Highlighting selected LINgroup is not available in SubmissionResult page.
• When clicking on a row in searchProcessView2, the page will proceed to GenomeProfile in the current tab. We need it be able to open that page in a new tab. Solved on 5/23/17
• Edit page seems to be wrong

Note

Front end

• Show LINgroup on result page
• Add “Type” column in result page for type strain: Y, N or N/A
• Add popup window on position headers to show thresholds

  Back end

• Have LINgroups associated to genomes in the database

  Collaboration

• Created a private github repository called CodeIgniter

Grant is leaving for his brand new start soon! Thanks a lot, it’s been great experience working with you!

This is the first LINbase blog post to take notes on to-dos of LINbase development.

Next you can update your site name, avatar and other options using the _config.yml file in the root of your repository (shown below).