-
Categories
-
Pharmaceutical Intermediates
-
Active Pharmaceutical Ingredients
-
Food Additives
- Industrial Coatings
- Agrochemicals
- Dyes and Pigments
- Surfactant
- Flavors and Fragrances
- Chemical Reagents
- Catalyst and Auxiliary
- Natural Products
- Inorganic Chemistry
-
Organic Chemistry
-
Biochemical Engineering
- Analytical Chemistry
- Cosmetic Ingredient
-
Pharmaceutical Intermediates
Promotion
ECHEMI Mall
Wholesale
Weekly Price
Exhibition
News
-
Trade Service
Written | In the process of research and treatment of prostate cancer, my best friend, Red Riding Hood, how the patient's genotype corresponds to the clinical phenotype is the focus and difficulty of the research in this field [1,2]
.
With the widespread application of machine learning in the field of biomedicine, it is possible to predict and discover potential cancer risks from the genome [3-5]
.
On September 22, 2021, the Eliezer M Van Allen research group from the Danner Faberge Cancer Research Center in the United States published an article titled Biologically informed deep neural network for prostate cancer discovery on Nature.
A deep machine learning model based on bioinformatics
.
Through this model, new treatment targets can be predicted for prostate cancer patients who are in a treatment-tolerant state
.
Specifically, based on the hierarchical concept of biological neural networks, the author integrated 3007 sets of biological network maps, constructed a deep learning prediction model, input the genome of prostate cancer patients, and through calculations, the first layer can output a series of gene clusters.
These Gene clusters can be connected by weighted lines; the second layer is related biological pathways; the next is to integrate the possible biological phenomena and conclusions predicted by these pathways, and so on, and finally the patient can be predicted ’S condition
.
The author collected data from 1013 prostate cancer patients, 80% of which were used for model training, 10% were used for model calibration, and the remaining 10% were used to test the accuracy of this model
.
From the results, the accuracy of the P-NET model built by the author is better than the known deep machine learning model under the condition of inputting the same volume of samples
.
Next, the author puts this model into practice
.
The author recruited two groups of prostate cancer patients, one group is in the primary state, and the other group has metastasized cancer lesions
.
The results show that the P-NET model can accurately predict 73% of the original state and 80% of the transferred state
.
The author guessed that the reason why a few patients in the primary state were classified as metastatic state by the P-NET model is probably because these patients are more severely ill and have a higher P-NET score
.
On the other hand, this model can not only predict the state of the patient, but also show the severity of the disease through the P-NET score
.
Next, in order to study the interaction between different genes, signal pathways, and biological networks and their influence on the model's prediction results, the author uses a schematic diagram to show the overall structure of the model's output (as shown in the figure below)
.
Among the many genomic changes in prostate cancer patients, the change in gene copy number has a greater impact on the disease than the gene mutation itself
.
In addition, in the 3007 group of signaling pathways covered by P-NET, multiple cell cycle-related pathways are related to the metastatic state, while the ubiquitination and SUMOylation signaling pathways are related to a variety of tumor suppressor factors, including AR, these pathways Abnormalities are likely to cause prostate cancer
.
For example, the bone-derived transcription factor RUNX2 can regulate the cell cycle and is associated with prostate cancer metastasis
.
Next, the author predicted a series of genes related to cancer progression through the P-NET model
.
The genes to focus on include AR, PTEN, RB1 and TP53, which are all known prostate cancer driver genes and have previously been reported to be associated with tumor metastasis
.
In addition, some genes, such as MDM4, FGFR1, NOTCH1 and PDGFA, are related to the accuracy of the prediction results
.
Finally, the author deeply analyzed the various levels of information output by the P-NET model and found that TP53-related pathways are related to prostate cancer, which has limited therapeutic effects of sexual organ removal
.
It has been reported that TP53 and the important gene MDM2 of its pathway are related to the course of prostate cancer.
The authors also found that the gene MDM4 is likely to have a similar effect
.
Previously, it has been known that MDM4 can inhibit the expression of TP53 by binding to the transcription region of TP53, but the relationship with the progression of prostate cancer is still not clear enough
.
In order to further study the specific role of the gene MDM4, the authors used CRISPR-Cas9 to specifically knock out MDM4 in a variety of prostate cancer cell lines
.
The results showed that compared with the control cells, the cell proliferation level of specifically knocked out MDM4 was significantly reduced
.
In addition, prostate cancer cell lines containing wild-type TP53 are more sensitive to MDM4 inhibitor RO-5963 than cell lines with TP53 mutations
.
In summary, the author gives a new method to solve biological problems by establishing a model based on deep machine learning through neural network mode
.
Moreover, the authors confirmed that with this model, the disease state of the patient can be predicted from the results of the gene sequencing of prostate cancer patients
.
In addition, based on the output information of the model, the author also predicted that some new targets, such as MDM4, are likely to be related to the course of prostate cancer, and confirmed by in vitro knockout methods that MDM4 deletion can inhibit the proliferation of prostate cancer cells
.
Furthermore, this model and related research methods are also expected to be extended to a variety of cancer types
.
Original link: https:// Platemaker: 11 References 1.
Robinson, D.
, et al.
Integrative clinical genomics of advanced prostate cancer.
Cell 161, 1215–1228 (2015).
2.
Abida, W.
, et al.
Genomic correlates of clinical outcome in advanced prostate cancer.
Proc.
Natl Acad.
Sci.
USA 116, 11428–11436 (2019).
3.
Ma, J.
, et al.
Using deep learning to model the hierarchical structure and function of a cell.
Nat Methods 15, 290–298 (2018).
4.
Yang, JH, et al.
A white-box machine learning approach for revealing antibiotic mechanisms of action.
Cell 177, 1649–1661.
e9 (2019).
5.
Kuenzi, BM, et al.
Predicting drug response and synergy using a deep learning model of human cancer cells.
Cancer Cell 38, 672–684.
e6 (2020) .
Reprinting instructions [original articles] BioArt original articles are welcome to be shared by individuals.
Reprinting is prohibited without permission.
The copyrights of all published works are owned by BioArt
.
BioArt reserves all legal rights, offenders must be investigated
.
.
With the widespread application of machine learning in the field of biomedicine, it is possible to predict and discover potential cancer risks from the genome [3-5]
.
On September 22, 2021, the Eliezer M Van Allen research group from the Danner Faberge Cancer Research Center in the United States published an article titled Biologically informed deep neural network for prostate cancer discovery on Nature.
A deep machine learning model based on bioinformatics
.
Through this model, new treatment targets can be predicted for prostate cancer patients who are in a treatment-tolerant state
.
Specifically, based on the hierarchical concept of biological neural networks, the author integrated 3007 sets of biological network maps, constructed a deep learning prediction model, input the genome of prostate cancer patients, and through calculations, the first layer can output a series of gene clusters.
These Gene clusters can be connected by weighted lines; the second layer is related biological pathways; the next is to integrate the possible biological phenomena and conclusions predicted by these pathways, and so on, and finally the patient can be predicted ’S condition
.
The author collected data from 1013 prostate cancer patients, 80% of which were used for model training, 10% were used for model calibration, and the remaining 10% were used to test the accuracy of this model
.
From the results, the accuracy of the P-NET model built by the author is better than the known deep machine learning model under the condition of inputting the same volume of samples
.
Next, the author puts this model into practice
.
The author recruited two groups of prostate cancer patients, one group is in the primary state, and the other group has metastasized cancer lesions
.
The results show that the P-NET model can accurately predict 73% of the original state and 80% of the transferred state
.
The author guessed that the reason why a few patients in the primary state were classified as metastatic state by the P-NET model is probably because these patients are more severely ill and have a higher P-NET score
.
On the other hand, this model can not only predict the state of the patient, but also show the severity of the disease through the P-NET score
.
Next, in order to study the interaction between different genes, signal pathways, and biological networks and their influence on the model's prediction results, the author uses a schematic diagram to show the overall structure of the model's output (as shown in the figure below)
.
Among the many genomic changes in prostate cancer patients, the change in gene copy number has a greater impact on the disease than the gene mutation itself
.
In addition, in the 3007 group of signaling pathways covered by P-NET, multiple cell cycle-related pathways are related to the metastatic state, while the ubiquitination and SUMOylation signaling pathways are related to a variety of tumor suppressor factors, including AR, these pathways Abnormalities are likely to cause prostate cancer
.
For example, the bone-derived transcription factor RUNX2 can regulate the cell cycle and is associated with prostate cancer metastasis
.
Next, the author predicted a series of genes related to cancer progression through the P-NET model
.
The genes to focus on include AR, PTEN, RB1 and TP53, which are all known prostate cancer driver genes and have previously been reported to be associated with tumor metastasis
.
In addition, some genes, such as MDM4, FGFR1, NOTCH1 and PDGFA, are related to the accuracy of the prediction results
.
Finally, the author deeply analyzed the various levels of information output by the P-NET model and found that TP53-related pathways are related to prostate cancer, which has limited therapeutic effects of sexual organ removal
.
It has been reported that TP53 and the important gene MDM2 of its pathway are related to the course of prostate cancer.
The authors also found that the gene MDM4 is likely to have a similar effect
.
Previously, it has been known that MDM4 can inhibit the expression of TP53 by binding to the transcription region of TP53, but the relationship with the progression of prostate cancer is still not clear enough
.
In order to further study the specific role of the gene MDM4, the authors used CRISPR-Cas9 to specifically knock out MDM4 in a variety of prostate cancer cell lines
.
The results showed that compared with the control cells, the cell proliferation level of specifically knocked out MDM4 was significantly reduced
.
In addition, prostate cancer cell lines containing wild-type TP53 are more sensitive to MDM4 inhibitor RO-5963 than cell lines with TP53 mutations
.
In summary, the author gives a new method to solve biological problems by establishing a model based on deep machine learning through neural network mode
.
Moreover, the authors confirmed that with this model, the disease state of the patient can be predicted from the results of the gene sequencing of prostate cancer patients
.
In addition, based on the output information of the model, the author also predicted that some new targets, such as MDM4, are likely to be related to the course of prostate cancer, and confirmed by in vitro knockout methods that MDM4 deletion can inhibit the proliferation of prostate cancer cells
.
Furthermore, this model and related research methods are also expected to be extended to a variety of cancer types
.
Original link: https:// Platemaker: 11 References 1.
Robinson, D.
, et al.
Integrative clinical genomics of advanced prostate cancer.
Cell 161, 1215–1228 (2015).
2.
Abida, W.
, et al.
Genomic correlates of clinical outcome in advanced prostate cancer.
Proc.
Natl Acad.
Sci.
USA 116, 11428–11436 (2019).
3.
Ma, J.
, et al.
Using deep learning to model the hierarchical structure and function of a cell.
Nat Methods 15, 290–298 (2018).
4.
Yang, JH, et al.
A white-box machine learning approach for revealing antibiotic mechanisms of action.
Cell 177, 1649–1661.
e9 (2019).
5.
Kuenzi, BM, et al.
Predicting drug response and synergy using a deep learning model of human cancer cells.
Cancer Cell 38, 672–684.
e6 (2020) .
Reprinting instructions [original articles] BioArt original articles are welcome to be shared by individuals.
Reprinting is prohibited without permission.
The copyrights of all published works are owned by BioArt
.
BioArt reserves all legal rights, offenders must be investigated
.
