CustOmics: A versatile deep-learning based strategy for multi-omics integration.

Hakim Benkirane, Yoann Pradat, Stefan Michiels, Paul-Henry Cournède

March 2023 PLoS Comput Biol

Synopsis of Social media discussions

The overall digital discussions reflect a strong interest in the technical innovation and applications of CustOmics, exemplified by comments highlighting the use of deep-learning strategies, interpretability features, and implications for cancer research. The tone varies from admiration to cautious optimism, with some participants referencing specific techniques like Shapley explanations and discussing the potential of the approach to advance multi-omics integration challenging traditional methods.

A
Agreement
 Moderate agreement

Most discussions acknowledge the significance of the CustOmics approach, viewing it as a valuable contribution to multi-omics integration, although some are cautious about its novelty.

I
Interest
 High level of interest

The posts demonstrate high interest, with frequent mentions of deep learning, multi-omics, and potential applications in biological research, indicating a keen engagement with the topic.

E
Engagement
 Moderate level of engagement

Posts reference technical aspects such as the model's architecture and interpretability methods, showing moderate to deep engagement rather than superficial mentions.

I
Impact
 Moderate level of impact

The discussions suggest that the methodology could influence future research, especially in cancer data analysis, indicating a moderate perceived impact.

Social Mentions

YouTube

1 Videos

Twitter

24 Posts

Metrics

Video Views

21

Total Likes

56

Extended Reach

147,371

Social Features

25

Timeline: Posts about article

Top Social Media Posts

Posts referencing the article

CustOmics: Deep Learning Strategy for Multi-Omics Data Integration

CustOmics: Deep Learning Strategy for Multi-Omics Data Integration

CustOmics is a deep learning approach for integrating various omics data types to enhance disease understanding and prediction accuracy. The model first trains on individual data sources, then learns their interactions, improving interpretability and performance in cancer classification and survival analysis.

Social RTCL TV

June 18, 2023

21 views

  • Valen Guerreros
    @GuerrerosValenn (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    April 5, 2023

    15

  • Lantiso
    @LittNFTs (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    April 4, 2023

    15

  • Sadad Redwan
    @SadadRedwan (Twitter)

    A versatile deep-learning based strategy for multi-omics integration #Learning via https://t.co/6qPcNWrkFh https://t.co/LLsR67EMZl
    view full post

    April 3, 2023

  • Benters Mall
    @BentersM (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    April 1, 2023

    15

  • nellister charles
    @NellisterC (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 30, 2023

    15

  • Stalwart
    @the_unswerving (Twitter)

    A versatile deep-learning based strategy for multi-omics integration #Learning via https://t.co/yNOfVDx5DP https://t.co/xo2bErz3Mc
    view full post

    March 25, 2023

  • S&M Telecommunications
    @sm_telecomm (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 22, 2023

    15

  • Antigoos
    @antigoos_art (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 21, 2023

    15

  • antisense.
    @razoralign (Twitter)

    CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43qgMT https://t.co/Ow5yV7G0pN
    view full post

    March 13, 2023

    2

  • Dmitry Svetlichnyy
    @dmitry_svetlich (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 8, 2023

    15

  • Kamal (卡馬爾) (@kamalmdm.bsky.social)
    @KamalMdM (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 8, 2023

    15

  • Hyungyong Kim
    @yong27 (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 8, 2023

    15

  • Saubashya Sur, PhD
    @SaubashyaSur (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 8, 2023

    15

  • In Absentia
    @tiwarko37 (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 8, 2023

    15

  • Toyota_Group_Komaba
    @KomaToug (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 7, 2023

    15

  • Seyoon Lee
    @Seyoon_L (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 7, 2023

    15

  • Xaira Rivera
    @xaairg (Twitter)

    RT @razoralign: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl…
    view full post

    March 7, 2023

    15

  • antisense.
    @razoralign (Twitter)

    CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/IkIk43pIXl https://t.co/tyFAfl5h1x
    view full post

    March 7, 2023

    46

    15

  • Oncology & Machine Learning
    @MlOncology (Twitter)

    CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/0o5UWaCQh1
    view full post

    March 6, 2023

  • Omkar
    @OmkarChandra (Twitter)

    RT @Deep__AI: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/j1CA5Nd7c5 by Hakim Benkirane et…
    view full post

    September 19, 2022

    2

  • arXiv Daily
    @Arxiv_Daily (Twitter)

    RT @Deep__AI: CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/j1CA5Nd7c5 by Hakim Benkirane et…
    view full post

    September 19, 2022

    2

  • DeepAI
    @DeepAI (Twitter)

    CustOmics: A versatile deep-learning based strategy for multi-omics integration https://t.co/j1CA5Nd7c5 by Hakim Benkirane et al. #DeepLearning #Autoencoder
    view full post

    September 18, 2022

    8

    2

  • Arxiv Bot
    @BotArxiv (Twitter)

    CustOmics: A versatile deep-learning based strategy for multi-omics integration #QuantitativeBio #bio #biology #arxiv https://t.co/wL7xb7qTsj
    view full post

    September 14, 2022

  • Machine Learning in Chemistry
    @ML_Chem (Twitter)

    CustOmics: A versatile deep-learning based strategy for multi-omics integration. (arXiv:2209.05485v1 [https://t.co/RzFGwRLevZ]) #machinelearning https://t.co/CAAalOFuyj
    view full post

    September 14, 2022

Abstract Synopsis

  • The paper discusses a new deep learning strategy called CustOmics for integrating multiple types of omics data (like genetic, protein, and other biological data) to better understand diseases and build predictive models.
  • It introduces a two-phase approach where the model first trains independently on each data source and then learns how these sources interact, leading to more efficient use of all data types and more accurate predictions, especially for cancer classification and survival prediction.
  • Additionally, the method emphasizes interpretability by adapting to Shapley additive explanations, allowing researchers to understand how different data sources contribute to the predictions, with performance demonstrated across various cancer datasets.]