Wednesday, July 27, 2022

a Sierpinski gasket has infinite perimeter

 


The length of the boundary of the nth iterate of the original triangle is the total length of the boundaries of all the shaded small triangles in the nth iterate. It can be shown that this gets arbitrarily large as n gets arbitrarily large. Therefore we conclude that a Sierpinski gasket has infinite perimeter!


alphafold firefly test run

 


[hqin@firefly ~]$ bash /opt/ohpc/pub/singularity/test-scripts/alphafold-2/alphafold-singularity-run.sh

Mounting /scr -> /mnt/data_dir

Mounting /scr/alphafold-data/uniref90 -> /mnt/uniref90_database_path

Mounting /scr/alphafold-data/mgnify -> /mnt/mgnify_database_path

Mounting /scr/alphafold-data/bfd -> /mnt/bfd_database_path

Mounting /scr/alphafold-data/uniclust30/uniclust30_2018_08 -> /mnt/uniclust30_database_path

Mounting /scr/alphafold-data/pdb70 -> /mnt/pdb70_database_path

Mounting /scr/alphafold-data/pdb_mmcif -> /mnt/template_mmcif_dir

Mounting /scr/alphafold-data/pdb_mmcif -> /mnt/obsolete_pdbs_path

--data_dir=/mnt/data_dir/alphafold-data --uniref90_database_path=/mnt/uniref90_database_path/uniref90.fasta --mgnify_database_path=/mnt/mgnify_database_path/mgy_clusters_2018_12.fa --bfd_database_path=/mnt/bfd_database_path/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt --uniclust30_database_path=/mnt/uniclust30_database_path/uniclust30_2018_08 --pdb70_database_path=/mnt/pdb70_database_path/pdb70 --template_mmcif_dir=/mnt/template_mmcif_dir/mmcif_files --obsolete_pdbs_path=/mnt/obsolete_pdbs_path/obsolete.dat --output_dir=/mnt/output --benchmark=0 --logtostderr --fasta_paths=/opt/ohpc/pub/singularity/test-scripts/alphafold-2/T1050.fasta --max_template_date=2021-12-31

/scr:/mnt/data_dir,/scr/alphafold-data/uniref90:/mnt/uniref90_database_path,/scr/alphafold-data/mgnify:/mnt/mgnify_database_path,/scr/alphafold-data/bfd:/mnt/bfd_database_path,/scr/alphafold-data/uniclust30/uniclust30_2018_08:/mnt/uniclust30_database_path,/scr/alphafold-data/pdb70:/mnt/pdb70_database_path,/scr/alphafold-data/pdb_mmcif:/mnt/template_mmcif_dir,/scr/alphafold-data/pdb_mmcif:/mnt/obsolete_pdbs_path,/home/hqin:/mnt/output


alphafold firefly

 



Singularity container environment is now available on Firefly. And I prepared an image for running Alphafold and designed a customizable run script for interacting with the instance.

To quickly test Alphafold with default config and input, run the following:
$ bash /opt/ohpc/pub/singularity/test-scripts/alphafold-2/alphafold-singularity-run.sh

Here's a list of locations for relevant directories:
- Alphafold inference data: /scr/alphafold-data
- Singularity image Alphafold: /opt/ohpc/pub/singularity/images/alphafold-2.sif
- Sample run script and fasta file: /opt/ohpc/pub/singularity/test-scripts/alphafold-2

Monday, July 25, 2022

deep learning on zhihu

 李沐的深度学习课

https://www.zhihu.com/education/video-course/1647604835598092705


Thursday, July 21, 2022

UTC MS graduate application


 

You have missed the fall deadline to submit and complete all requirements.  Spring 2023 would be your best bet.  Files are not reviewed for an admission decision until all academic requirements have been submitted.

 

The following requirements must be fulfilled by July 1st for the fall terms and by November 1st for the spring terms 

 

Please note, we do not begin new students in the summer term due to immigration requirements.

 

Refer to the programs link below for each department’s requirements.

 

  1. Submit the online International graduate application for either fall or spring.  We do not begin new international students in the summer term unless the program only begins in the summer.
  2. Pay the $40.00 application fee.
  3. Submit official transcripts and degree certificates from each college or university you have attended.
  4. See graduate programs offered at UTC and the program requirements. Some programs require other test scores.  Click the link to see a list of programs we offer and the programs requirements and deadlines.
  5. Report satisfactory English test scores from one of the three test scores (institution code 1831). TOEFL IBT minimum score: 79.  IELTS minimum score: 6. Duolingo minimum score 100. MS Computer Science requires TOEFL_ 83ibt, IELTS 6.5 and Duolingo107.
  6. Institutional code is 1831 for reporting test scores.

Submitting official proof of funding can be done after the file has been reviewed for an admission decision but before the deadline.

Generally, International students are not eligible for graduate assistantships during the first semester at UTC.  Upon completion of one full-time semester and a GPA cum of 3.25 or higher, you may be considered for graduate assistantship.  It is important to note graduate assistantships are highly competitive and based on merit, not monetary need.

IF a GA is available, they are given by the program to which you applied and accepted, not by our office.

Visit www.utc.edu/international for more information regarding housing, orientation, student fees, health insurance and the admissions process.

If you have any questions, please don’t hesitate to email me.


Tuesday, July 19, 2022

youth’ protein may drive aging in the eye

 

NIH study finds loss of ‘youth’ protein may drive aging in the eye

Loss of the protein pigment epithelium-derived factor (PEDF), which protects retinal support cells, may drive age-related changes in the retina, according to a new study in mice from the National Eye Institute (NEI). The retina is the light-sensitive tissue at the back of the eye, and aging-associated diseases of the retina, like age-related macular degeneration (AMD), can lead to blindness. This new finding could lead to therapies to prevent AMD and other aging conditions of the retina. The study was published in the International Journal of Molecular Sciences. NEI is part of the National Institutes of Health.

Reference:
https://www.nih.gov/news-events/news-releases/nih-study-finds-loss-youth-protein-may-drive-aging-eye

Rebustini IT, Crawford SE, Becerra SP. “PEDF deletion induces senescence and defects in phagocytosis in the RPE.” July 13 2022. Int J Mol Sci. https://doi.org/10.3390/ijms23147745



nucleotide symbol


Nucleotide symbols

Nucleotide symbolFull Name
AAdenine
CCytosine
GGuanine
TThymine
UUracil
RGuanine / Adenine (purine)
YCytosine / Thymine (pyrimidine)
KGuanine / Thymine
MAdenine / Cytosine
SGuanine / Cytosine
WAdenine / Thymine
BGuanine / Thymine / Cytosine
DGuanine / Adenine / Thymine
HAdenine / Cytosine / Thymine
VGuanine / Cytosine / Adenine
NAdenine / Guanine / Cytosine / Thymine

 





geometry, polyhedra


 



UTC fee and tuition

 https://www.utc.edu/finance-and-administration/office-of-bursar/fee-information/fall-2022-fee-schedules


instate

https://www.utc.edu/sites/default/files/2022-06/FY%202023%20-%20In-State%20Fee%20Schedule.pdf


out of state


Monday, July 18, 2022

How to Train Really Large Models on Many GPUs? September 24, 2021 · 21 min · Lilian Weng

 

How to Train Really Large Models on Many GPUs?

https://lilianweng.github.io/posts/2021-09-25-train-large/


Are Deep Neural Networks Dramatically Overfitted? March 14, 2019 · 22 min · Lilian Weng

 https://lilianweng.github.io/posts/2019-03-14-overfit/

Are Deep Neural Networks Dramatically Overfitted?

Sunday, July 17, 2022

GISAID alignment data

 

The msa alignment in GISAID in 2022 is only limited to the reference sequence. So, the position is in the alignment should match the reference genome. 

GISAID reference wuhan-hu-4

hCoV-19/Wuhan/WIV04/2019|EPI_ISL_402124, full length 29891 nucleotides. 



rotation matrix

 rotation matrix


https://en.wikipedia.org/wiki/Rotation_matrix 

Saturday, July 16, 2022

Friday, July 15, 2022

Species distribution models with BART

 

Species distribution models with BART

https://github.com/cjcarlson/embarcadero

Colin J. Carlson

Wednesday, July 13, 2022

Canadian Institute for Cybersecurity: Data sets

 

https://www.unb.ca/cic/datasets/index.html



a-tale-of-two-covariates-why-owid-and-company-are-wrong-about-us-healthcare/#rcatoc-notes

Healthcare and diminishing returns 

https://randomcriticalanalysis.com/2019/11/07/a-tale-of-two-covariates-why-owid-and-company-are-wrong-about-us-healthcare/#rcatoc-notes



Tuesday, July 12, 2022

predicting new viral variants

 

Flagship unwraps new AI biotech that looks to predict variants before they’re here


https://endpts.com/flagship-unwraps-new-ai-biotech-that-looks-to-predict-variants-before-theyre-here/


Friday, July 8, 2022

ethics in AI

Q1: What are potential sources of bias that may explain NIST’s findings? Should there be

limitations placed on the use of facial recognition technology based on your assessment

of NIST’s study?


Source of images: “All came from operational databases provided by the State Department, the

Department of Homeland Security and the FBI.”


Uneven Training data? 

Unequal sampling?



Q2: Facial recognition technology is becoming ubiquitous to access phones, computers, US

taxes, etc. For the application of facial recognition technology in healthcare, what are

some unique considerations that developers/policymakers/scientists should

acknowledge/apply?


Data privacy?


Potential data theft 




Q3: To what extent are patients, providers (e.g., clinicians, hospitals, health systems), payers

(e.g., insurers, employers), and policymakers (e.g., healthcare and insurance regulators,

state Medicaid directors) aware of the inclusion of variables based on race/ethnicity in

healthcare algorithms and algorithm-informed decision tools?


Payment issues


 


Monday, July 4, 2022

multinomial growth fitness model

 

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1086494/Technical-Briefing-43-28.06.22.pdf


Friday, July 1, 2022

BA substrain comparison

Using Qin's cumulative N ratio method 

BA.2 / BA.1, slope = 0.01464467


BA.4 / BA.2, slope = 0.02896763

BA.5 / BA.1

BA.4 / BA.1


















biomedical ML/AI

 lead students collectively write a survey paper on github
This course will discuss the research forefronts and breakthroughs of artificial intelligence in the field of biomedical related fields, including highly accurate protein structure prediction with AlphaFold, fast and energy-efficient neuromorphic deep learning with first-spike times, machine learning platform to estimate anti-SARS-CoV-2 activities, adversarial interference and its mitigations in privacy-preserving collaborative machine learning; machine learning and algorithm fairness in public and population health, and computer vision in healthcare

nature machine learning
Aviv Regev works
CSHL meeting talks
pipp workshop reports
https://www.cc.gatech.edu/~badityap/ 
https://www.biorxiv.org/content/10.1101/803205v2#readcube-epdf
https://www.nature.com/natmachintell/research-articles
Navigating the pitfalls of applying machine learning in genomics
https://www.nature.com/articles/s41576-021-00434-9 

Collection of ML/AI pitfall papers
https://github.com/crazyhottommy/machine-learning-resource/blob/master/README.md