Discoveries Weekly No. 11 (June 22-28, 2020)
Here are some discoveries that fascinate me this week.
I recorded this week Sonatina by James Hook (1746-1827), an English composer and organist, with the indication Allegro non troppo, which means fast, but not too fast. In case you like the recording, please feel free to share it.
Biology
Accelerating Cancer Immunotherapy Research (ACIR)
I encountered a website that brings researchers to the latest development of cancer immune therapy, the ACIR. It aggregates latest publications and research news about (cancer) immunology and summarizes them in compact articles. I recommend it to people who are interested in the topic.
For instance, TOX is so exhausting, a post in July 2019, summarizes five publications about the mechanism how transcription factor TOX (thymocyte selection associated high mobility group box) contributes to the formation of exhausted CD8+ T cells via chromatin remodelling and alteration of RNA transcriptome. I learned a lot about T cell exhaustion in cancer immunotherapy by reading the summary.
Intrinsically disordered regions in transcription factors
See my learning notes.
Other gems in biology
- Starr et al. around Jesse Bloom at Fred Hutchinson Cancer Research Center reported a library of single-amino acid mutants of the ACE2 receptor binding region of the SARS-nCov-2 virus. It was also covered by Derek Lowe’s blog post A Wide Look at Coronavirus Mutants.
- Young blood and old blood in the In The Pipeline blog introduced two recent publications about attempts to reset genome methylation to turn back the aging clock (coming from a start-up and people in Horvath’s group at UCLA), referring to the preprint Reversing age: dual species measurement of epigenetic age with a single clock.
- Much research seems being done in the field of aging. Amor et al. reported on Nature their use of chimeric antigen receptor (CAR) T cells to target senescent cells, using the urokinase-type plasminogen activator receptor (uPAR, gene symbol PLAUR) as a cell-surface protein marker of senescent cells. They showed elongated survival in a mouse model of lung adenocarcinoma, and improved homeostasis in a mouse model of liver fibrosis.
Computational biology
Evaluating single-cell structure preservation by dimensionality reduction
Heiser and Lau evaluated dimension-reduction techniques. They showed that input cell distribution (the biology of the cells, I would say) is the most dominant factor, followed by method parameters. My takeaway, though not surprisingly, is that the applicability is technology- and data-specific, and there are no universally applicable models. Among the tested methods, UMAP and t-SNE, two commonly used methods, give very similar results.
Mathematical modelling and single-cell RNA sequencing for cancer immunotherapy
Please find my learning notes in another blog post about an interesting study by Griffiths et al. that combines mathematical modelling and single-cell RNA sequencing to understand patient response to cancer immunotherapy.
Mathematical oncology
I discovered an interesting blog and email list on mathematical oncology, Mathematical Oncology, created by Jeffery West and colleagues. It sends weekly emails about progress in mathematical oncology (RSS is also possible). Given the large amount of data and relative good funding of oncology compared with other disease indications, the modelling approach there can be interesting for other disease areas as well.
Programming
Jupyter nbconvert
The following command can be used to run a Jupyter notebook from the command line, making changes in place.
jupyter nbconvert --ExecutePreprocessor.timeout=600 --to notebook
--inplace --execute standard_workflow_besca2.0.ipynb
The --ExecutePreprocessor.timeout=600 option tells the program that each cell
can run at maximum 600 seconds. The output will be written in the notebook (--
to notebook --inplace).
More can learned from the execute API of Jupyter nbconvert.
Other gems
Researchers and founders
The blog post by Sam Altman summarizes his experience working with researcher and founders. A relevant blog post by John Schulman reflects some habits shared by successful researchers in the field of machine learning.
Some key points that I took:
- The most productive founders and researchers keep pondering upon the Hamming’s question: ‘what are the most important problems in your field, and why aren’t you working on them?’ Honing the taste about problems is an important skill that we need to learn all the time.
- An effective researcher or founder needs both laser-sharp focus upon what is coming next and long-term vision. There can be ‘idea-driven’ or ‘goal-driven’ approach to research. I agree with John Schulman that ‘goal-driven’ approach may be a more practical one for people working in a team.
- Being persistent and working hard makes it more likely to be successful. I wonder how that can be combined with a role in the family and a role for the society. Maybe the answer is that we have to work as much as our physical and mental health, our family, and our social need can support.
- Allocate time for broad-scope learning and personal development.
- Effective people bias towards action and trying things, and being honest about what works and what does not work.
- Effective people are more creative and generating more ideas.
- Productive people value autonomy, and are not confined by the rules that make little sense for them.
- The motivations can be complex - often driven by genuine curiosity.
DNA surveillance in China
According to New York Times, China is collecting DNA information from 10% of men and boys. The topic was also covered by Nature.
Predicting mortality
Is death a predictable event? It seems that for some degree, it may be. Puterman et al. reported on PNAS their findings by building statistical models of mortality using more than 50 factors.
The following figure (Figure 3) tells much of the story. The higher the hazard ratio, the more likely that the factor causes death.
If we take the correlation as causality, which can be totally wrong, here are the things that we can do to ourselves and to others if the goal is to live longer (that may not be the only goal of life though):
- Do not smoke. Never.
- If possible, do not drink alcohol. If impossible, do it with caution.
- Exercise.
- Sleep well.
- Get education.
- Plan your career, finance and wealth (it may not be in our control, though).
- Know your neighbours.
- Spend time with spouse, friends, and family you cherish.
- Find a purpose in life (that takes a life, boy, I was told).
- Become satisfied with what you get in life.
- Stay conscientious, namely being careful, diligent (hard-working and focused), is about as important as being optimistic.
- Give ourselves and others hope, not discrimination.
It sounds almost mundane, right? The conclusions are largely consistent the other study about happiness and sadness. There may be more things that the numbers and the statistical models can tell us. And it helps if we can quantify the contribution of individual factors as the authors of the paper did, despite that some factors are strongly correlated via latent variables. The real challenge is how to live up to them as often and as well as we can, given events and changes in life. That is a game that restarts everyday.
Turtles and tortoises are in trouble
Turtles and tortoises are in trouble is a review authored by Stanford et al. and published in Current Biology. It sheds light on the urgent extinction pressure of turtles (which mainly live in waters) and tortoises (which mainly live on the land), which are known collectively as chelonians. The bleak prediction goes that many species will go extinct this century.
Save the earth (and turtles and tortoises) before it is too late while enjoying the weekend!