Autumn 2021 Update
SemiBin: A better metagenomics binner. Also, we're at #MVIF2021!
Autumn 2021 Update
This is our third quarterly update. We will send four of these a year, (roughly) coinciding with the two equinoxes and the two solstices. Every four months, we highlight one of the projects in the group and give short updates on everything else that has been happening.
This time the focus is on Ph.D. student Shaojun Pan and his work on a recently released binning tool – SemiBin. Below you can find some exciting BDB-Lab news and the invitation to join the Microbiome Virtual International Forum!
Autumn 2021 Focus: SemiBin by Shaojun Pan
What are MAGs or population genomes? Why are they important?
A metagenome-assembled genome (MAG) is a genome obtained from metagenomics through computational analysis. As many microorganisms present in the environment do not have a genome in public databases, we can use these MAGs to analyze their function and the composition of the microbiome as a whole.
What is SemiBin?
SemiBin (semi-supervised metagenomic binning) is a metagenomic binning tool implemented in Python. You can easily install it with bioconda.
What can be done with SemiBin that makes it different from other state-of-art programs?
SemiBin is a semi-supervised binning tool which uses siamese neural networks to learn information from reference genomes and can reconstruct genomes beyond the reference datasets. We demonstrated that the use of deep learning could improve binning results, outperforming existing binning tools in both simulated and real datasets.
Where do you plan to take the project in the future?
We want to take advantage of big data in metagenomics to get better models and even better binning results.
Can you also tell us a bit about yourself: what was your path to get here?
I have a computer science background. Now I am a Ph.D. student majoring in biomedical engineering. I was very interested in using machine learning and deep learning methods to solve metagenomic questions.
What are your future (scientific) plans?
I have a long term plan to keep extending the SemiBin machine learning model to incorporate more ideas from what is called self-supervised learning.
Right now, I am also working on making SemiBin more usable and spreading its use. For example, on November 2nd, we will have an online tutorial (sign up now).
Where can people find you and get in touch?
My email is shaojun1997777@gmail.com.
BDB-Lab Updates
People. This summer we had three remote interns joining us: Anna Vines, Ariana Thakurdyal, and Nilesh Gupta. Their projects were successfully concluded and their blog posts will be up on our website soon. Ariana Thakurdyal presented her project "Health risk assessment of global resistome" at the NSURP 2021 on the final presentation day (recording available here (46:22)).
Anna Cuscó recently joined BDB-Lab as a postdoctoral researcher working on our global microbiome analyses. Hui Chong also joined us as a research assistant and is collaborating with several projects in our group. To know them a bit better, visit our page of members.
Manuscript. The SemiBin manuscript (our innovative binning tool) was made public as a preprint and was submitted for publication.
Tutorial. The tutorials for NGLess and Jug were concluded. These were online hands-on sessions to train people on using our tools and attracted the interest of a lot of people. Additionally, these events provide valuable feedback about our tools, which we can use to enhance usability.
On November 2nd (noon UTC) we will run a SemiBin tutorial (sign-up now). We intend to announce more tutorials in the near future. Sign up for the tutorials mailing-list to keep informed.
Online resource. Besides microbiome events, Svetlana is publicly curating a list of podcasts related to microbiome science. If you want to share your event or podcast, feel free to let her know.
Microbiome Forum. Luis and Svetlana joined the steering committee of the Microbiome Virtual International Forum. MVIF is a virtual, efficient, and inclusive alternative to a traditional multi-day microbiome conference. This free, monthly event is available for Atlantic and Pacific time zones, aiming to gather microbiome scientists from all around the world. We hope to see you there!
Looking Forward
Svetlana will take part in the 7th Microbial Bioinformatics Hackathon with a special focus on Antimicrobial Resistance (October 11-13, 2021) organized by the Public Health Alliance for Genomic Epidemiology (PHA4GE), Joint Programming Initiative on Antimicrobial Resistance (JPIAMR), and Cloud Infrastructure for Big Data Microbial Bioinformatics (CLIMB-BIG-DATA project).
Shaojun submitted an abstract to present SemiBin at the Microbiome Virtual International Forum.