When genetic sequencing can save lives, how can it be slow?

In 2013, Hollywood superstar Angelina Jolie went for genetic testing because she was worried about the risk of breast cancer inherited in her family, and it turned out that she carried a mutation in the BRCA1 gene, which gave her an 87% risk of breast cancer and a 50%
risk of ovarian cancer.
To avoid suffering from cancer at the age of forty or fifty like her mother and grandmother, Julie opted for mastectomy, and in 2015, she surgically removed her ovaries and fallopian tubes
.
She believes genetic testing has saved her life
.

Julie was able to make this choice thanks to the important technology
of gene sequencing.
Before we can understand gene sequencing technology, we must first understand what genes are
.

Genes and gene sequencing

As early as 1909, Danish geneticist W.
Johansen officially proposed the concept
of "gene".
In 1953, James Watson and Francis Crick proposed the double helix structure
of DNA.
Since then, people have begun to understand the nature of genes - DNA fragments
with genetic effects.
These DNA fragments are like pieces of code, recording and storing all the information about the
species, race, blood type, gestation, growth, apoptosis and other processes of life.

As small as bacteria, as large as blue whales, and as high as humans, genes composed of DNA are passed down from generation to generation, and they constitute the "genetic code"
of life.
For us humans, why we become the only intelligent beings, our height, skin color, hair color, intelligence, various genetic diseases, cancer risk, lifespan, etc.
, are closely related
to genes.
In short, the birth, growth, decline, disease, aging, death and other life phenomena of organisms are largely determined
by genes.

In order to uncover the secret of genes, in the 90s of the last century, under the promotion of scientists all over the world, the "Human Genome Project" with the same name as the "Manhattan Project" and the "Apollo Program" was officially launched
.
The project hopes to genetically sequence 3 billion DNA sequences across the human genome to help us better understand humans themselves
.

In 2001, after more than a decade and nearly $3 billion in funding, a draft of the human genome work was released, marking the success of the Human Genome Project, and mankind understood and mastered its genetic code
for the first time.
This is not only a great scientific project, but also lays the foundation
for us to decode life, understand the origin of life, growth and development, understand diseases, treat diseases, and fight aging.

However, the success of the Human Genome Project is only the beginning
.
We have found only more than 20,000 protein-coding genes from the human genome map, which account for less than 3% of the human genome, and the vast majority of the rest are non-coding sequences that do not code for proteins, which remain hidden in the dark, waiting for scientists to explore
.

What are the other difficulties in gene sequencing?

More than 20 years ago, the Human Genome Project spent nearly $3 billion to complete the sequencing of most of the human gene sequences, which is obviously not affordable for conventional scientific research, let alone consumer-grade gene sequencing
.
With advances in sequencing technology, the cost of sequencing genes has dropped rapidly, and the cost of sequencing a person's entire genome has now fallen below $1,000
.

However, whether for scientific research or consumer-grade gene sequencing, the data generated in the gene sequencing process is very large (personal genome raw data can reach 1TB, and the Million People Genome Project involves exabyte-level data processing), which requires high
computing power.
The sequencing analysis process is complex and has a low degree of automation, which requires a lot of manpower and time to repeat the development process, resulting in low
efficiency.
These difficulties hinder the further reduction of the cost of gene sequencing, and also hinder the application of
scientific research and consumer-grade gene sequencing.

HUAWEI CLOUD AI Gene Platform provides one-stop support for genetic analysis

1.
The help of EIHealth AI gene platform

Based on the advantages of HUAWEI CLOUD AI and big data, HUAWEI CLOUD has launched a medical intelligent agent (EIHealth) to provide a professional AI R&D platform
for genomic analysis, drug R&D, and clinical research.
The HUAWEI CLOUD EIHealth AI Gene Platform supports the implementation
of gene scenario services by helping to solve massive data storage, elastic scaling of resources, reducing duplicate development, and multi-project/multi-user collaboration.

Specifically, in large-scale production scenarios, the AI gene platform can support exabyte-level storage, which is not limited by hard disks, meet the storage of massive data, and save labor costs through batch automation, thereby helping to reduce resource costs by 30%.

When the business has tidal attributes, the AI gene platform can realize the elastic expansion and contraction of resources, and the business is concentrated in 1/3 time, which can help save more than 40% of the cost
.

In the face of organizational collaboration scenarios, the AI gene platform can realize resource sharing within the organization, reduce repeated development, increase personnel collaboration efficiency by 30%, reduce labor development costs by 50%, and increase asset reuse efficiency by 100%.

At the same time, the AI gene platform also has AI capabilities
.
Experts in the field of biomedicine generally have relatively weak AI algorithm capabilities, and most of them use traditional statistical algorithms
.
HUAWEI CLOUD AI Genomics Platform can automatically model, analyze, and interpret genomic data through the unique AutoGenome algorithm, helping to automatically find parameters and automatically conduct interpretability studies
on models.
Users only need to configure 5 lines of code to build high-precision AI gene models, covering monoomics, multi-omics and gene action networks
.

In terms of performance, modeling performance for genomes is generally improved by 10%, and biomarkers can be found
.
This is very important for biomedical customers, which means that the model is not only accurate, but also knows why it is accurate, and then just look at this biomarker, you can know what drugs and tests
the patient should take.

It is worth mentioning that after the completion of gene sequencing, due to the large amount of data, the delivery method of analysis results is often mailed to a mobile hard disk, which leads to long lead times and high
costs of results.
The AI gene platform can open sub-accounts to customers, and customers can directly log in and download them, saving hard disk costs and mailing costs, and shortening the delivery cycle
.

2.
Case sharing

In foreign countries, HUAWEI CLOUD supports the Million People Genome Project, which analyzes more than 1 petabyte of data in a single day, which is more than 10 times faster than traditional HPC (high-performance computing) solutions, and reduces costs by 30%.

In China, clinical researchers of Peking Union Medical College Hospital discovered lung cancer pharmacodynamic markers based on the EIHealth platform, which provided a new perspective for the classification of clinical response of mEGFR lung adenocarcinoma patients, and revealed the potential of
non-invasive pretreatment of serum metabolites in predicting the efficacy of EGFR-TKI.
The study was published in the American Journal of Cancer Research [1].

Image

3.
Explore non-coding sequences

As we mentioned earlier, there are only over 20,000 genes in the human genome that code for proteins, and as of now, almost all drugs target more than 700 of these more than 20,000 proteins associated with disease
.
This is obviously too little, and it has led to many diseases that are incurable
.
Gene sequencing and analysis of more than 97% of the DNA sequences in the human genome is non-coding, a blue ocean of drug target discovery, which helps us redraw the human genome and greatly expand the number of
potential therapeutic targets.

What else can AI do in the medical field

In 2016, Google released its first AI medical algorithm[2], which can detect diabetic retinopathy from retinal fundus photos through AI, with specificity and sensitivity comparable to that of a professional ophthalmologist
.
The prelude
to AI+ medical treatment began.
Since then, pathological diagnosis and auxiliary treatment products based on artificial intelligence (AI) have shown a vigorous development trend
as a whole.
In addition, AI pharmaceuticals, AI protein structure prediction and other fields are also advancing by leaps and bounds, and AI is changing the entire biomedical field
with unprecedented speed and efficiency.

HUAWEI CLOUD facilitates the discovery of super antimicrobials

In February 2022, Professor Liu Bing's team from the First Affiliated Hospital of Xi'an Jiaotong University published a research paper in the internationally renowned academic journal Proceedings of the National Academy of Sciences (PNAS) [3].

Starting from the bacteriophage, the natural enemy of bacteria, Professor Liu Bing's team discovered for the first time in the world that bacteriophage encodes a bacteriostatic gene that targets bacterial histone HU, and elaborated on the detailed mechanism of bacteriophage Gp46 protein inhibiting bacteria by binding to HU, and on this basis, developed a new class of super antibacterial drugs, Drug X, which is expected to become the world's first new target and new class of antibiotics
in the past 40 years.

Professor Liu Bing used the AI-assisted drug design service based on the HUAWEI CLOUD Pangu drug molecule model to find a breakthrough in finding small molecule compounds that can replace Gp46 protein, developing the super antibacterial drug Drug X, and breaking the "Double Ten Law"
in the pharmaceutical industry.
Professor Liu Bing said that the HUAWEI CLOUD Pangu Drug Molecule Model has shortened the R&D cycle of lead compounds from several years to one month, and reduced R&D costs by 70%.

Image

Drug molecules screened based on HUAWEI CLOUD Pangu drug molecule model

Respiratory chronic disease + AI, improve quality control accuracy and discrimination accuracy

Respiratory chronic diseases are the third most common chronic diseases
in China after hypertension and diabetes.
The patient base is large, and the ability of hospitals at all levels to apply pulmonary function tests and quality assurance is uneven, especially in grassroots hospitals
.
To improve the screening and management capabilities of chronic respiratory diseases at the grassroots level, HUAWEI CLOUD EI Innovation Incubation Lab, together with partners such as the National Respiratory Center of the First Hospital of Guangzhou Medical Hospital, Yiyoulian, and Saike, jointly developed a big data and AI-assisted system
for lung function.
According to large-scale test results, HUAWEI CLOUD AI has steadily improved the quality control accuracy and discrimination accuracy of respiratory chronic disease detection, from 50% to 90%, which is higher than the average level
of doctors surveyed in 90 hospitals.

AI auxiliary diagnosis of cardiovascular and cerebrovascular diseases, continuous breakthroughs in various fields

Cerebral aneurysms rank among the top 3 causes of cerebrovascular diseases, which can be called "silent and deadly killers"
.
In response to the problem of auxiliary diagnosis of aneurysm, HUAWEI CLOUD EI Innovation Incubation Lab, together with the School of Telecommunications of Huazhong University of Science and Technology and the Department of Radiology of Tongji Medical College Affiliated Union Hospital of Huazhong University of Science and Technology, used the ModelArts development platform of HUAWEI CLOUD AI development line to develop a fully automated and highly sensitive brain aneurysm detection algorithm based on CTA images, with a sensitivity of up to 97.
5%, helping doctors improve the sensitivity of clinical diagnosis by about 10 percentage points and reduce the missed diagnosis rate by 5 percentage points.
At the same time, it effectively shortens the doctor's diagnosis time
.

AI-assisted cervical cancer screening model promotes early screening and treatment

In the field of cervical cancer screening, the HUAWEI CLOUD AI team worked with Jinyu Medical Pathology experts to train an accurate and efficient AI-assisted cervical cancer screening model
based on pathological morphology and deep learning technology.
On the basis of the negative exclusion rate higher than 60%, the correct rate of negative film interpretation is higher than 99%, and the detection rate of positive lesions is more than 99.
9%.

At the same time, each pathological interpretation only takes 36 seconds, and the interpretation speed is 10 times
that of manual interpretation.

The AI-assisted cervical cancer screening method has special value in areas with few medical resources, which will greatly improve the population coverage and service frequency of cervical cancer screening services, and promote early screening and treatment
of cervical cancer.

The infinite possibilities of AI+ life sciences

In recent years, AlphaFold, an artificial intelligence program developed by DeepMind, has achieved accurate prediction of protein structure, allowing the world to see the powerful potential
of artificial intelligence (AI) in life sciences.
Today, AlphaFold2 and Meta AI predict almost all the billions of protein structures on Earth; RoseTTAFold, developed by David Baker's team at the University of Washington, is able to conceive and engineer proteins
through AI.

The HUAWEI CLOUD AI Gene Platform based on AI and big data provides high-performance, reliable, and cost-effective gene sequencing computation, storage, analysis, and AI capability support, standardizing and executing the scientific research process, thereby helping to speed up gene sequencing, improve efficiency, and reduce costs.
In addition, HUAWEI CLOUD platform can help pharmaceutical companies complete drug R&D faster and more efficiently, saving R&D costs.
It can also help clinical research, provide medical clinical data AI analysis model, patient group marker discovery service, drug efficacy prediction service, medical image big data intelligent labeling and AI-assisted diagnosis
.

AI is promoting the rapid development of life sciences at an unprecedented speed, AI pathology imaging, AI disease screening, AI pharmaceutical, AI gene analysis, AI medical robots, AI protein structure prediction and design, are all milestones on the way forward, the combination of AI and life sciences, there are more surprises waiting for us
.

Through AI for Healthcare, HUAWEI CLOUD will continue to contribute to human disease prevention, diagnosis and treatment, and drug research and development
.

References:

1.
https：//pubmed.
ncbi.
nlm.
nih.
gov/33414999/

2.
https：//jamanetwork.
com/journals/jama/fullarticle/2588763

3.
https：//www.
pnas.
org/doi/full/10.
1073/pnas.
2116278119

4.
https：//www.
nature.
com/articles/s41587-022-01221-5