Recent Advances on Penalized Regression Models for Biological Data

Wang, Pei; Chen, Shunjie; Yang, Sijia

doi:10.3390/math10193695

Cited by 5 publications

(4 citation statements)

References 134 publications

(238 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The same dataset was trained in elastic net regression analysis to confirm the predicted markers in PCA and informed decision-making against the dependent variable by score. The elastic net regression model demonstrated robust performance in handling multicollinearity and selecting relevant predictors, leading to improved predictive accuracy compared to traditional regression techniques [58]. The objective was to identify the combination that minimizes the model's error metrics, such as mean squared error (MSE) or R squared.…”

Section: Discussionmentioning

confidence: 99%

“…The correlations were interpreted based on the guidelines provided by Hinkle et al in 2003 [60], categorizing correlations as very high positive (negative) correlation (±0.90 to 1.00), high positive (negative) correlation (±0.70 to 0.90), moderately positive (negative) correlation (±0.50 to 0.70), low positive The same dataset was trained in elastic net regression analysis to confirm the predicted markers in PCA and informed decision-making against the dependent variable by score. The elastic net regression model demonstrated robust performance in handling multicollinearity and selecting relevant predictors, leading to improved predictive accuracy compared to traditional regression techniques [58]. The objective was to identify the combination that minimizes the model's error metrics, such as mean squared error (MSE) or R squared.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

The Effects of Artificial Diets on the Expression of Molecular Marker Genes Related to Honey Bee Health

Frunze,

Kim,

Lee

et al. 2024

IJMS

View full text Add to dashboard Cite

Honey bees are commonly used to study metabolic processes, yet the molecular mechanisms underlying nutrient transformation, particularly proteins and their effects on development, health, and diseases, still evoke varying opinions among researchers. To address this gap, we investigated the digestibility and transformation of water-soluble proteins from four artificial diets in long-lived honey bee populations (Apis mellifera ligustica), alongside their impact on metabolism and DWV relative expression ratio, using transcriptomic and protein quantification methods. Diet 2, characterized by its high protein content and digestibility, was selected for further analysis from the other studied diets. Subsequently, machine learning was employed to identify six diet-related molecular markers: SOD1, Trxr1, defensin2, JHAMT, TOR1, and vg. The expression levels of these markers were found to resemble those of honey bees who were fed with Diet 2 and bee bread, renowned as the best natural food. Notably, honey bees exhibiting chalkbrood symptoms (Control-N) responded differently to the diet, underscoring the unique nutritional effects on health-deficient bees. Additionally, we proposed a molecular model to elucidate the transition of long-lived honey bees from diapause to development, induced by nutrition. These findings carry implications for nutritional research and beekeeping, underscoring the vital role of honey bees in agriculture.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

The Effects of Artificial Diets on the Expression of Molecular Marker Genes Related to Honey Bee Health

Frunze,

Kim,

Lee

et al. 2024

IJMS

View full text Add to dashboard Cite

show abstract

“…[16][17][18] It offers a distinct advantage in mitigating the risk of over-fitting and reducing model variance through simultaneous feature selection and regularization. 19 LASSO proves particularly valuable when dealing with problems that involve numerous features (the extensive proteomics data per sample) but a relatively small sample size (the limited number of participants that meet the recruitment criteria for a cohort study). 20 In this cohort study, we assessed the neurological recoveries of patients who underwent intramedullary surgery, monitored changes in CSF protein profiles in these patients using TMT-MS analysis, and employed LASSO penalized regression modeling for identification of CSF protein sets associated with spinal function recovery in the corresponding patient populations.…”

Section: Introductionmentioning

confidence: 99%

Identification of Cerebrospinal Fluid Proteomic Signatures Associated with Intramedullary Surgery Outcome

Peng

et al. 2023

Preprint

View full text Add to dashboard Cite

Importance Neurological function recovery after intramedullary microsurgery is susceptible to various factors. Investigating alterations in cerebrospinal fluid (CSF) protein profiles associated with spinal function recovery could provide valuable insights into prognosis and the development of postoperative therapeutic approaches. Objective The aim of this research was to identify CSF protein sets linked to recovery outcomes of intramedullary microsurgery. Design This research is an observational cohort investigation that collected data from August 2020 to March 2023. Setting The Neurospine group conducted this single-centered cohort research (http://www.chictr.org.cn, registration ID: ChiCTR2300072704). Participants were followed up for 6 months or more. CSF samples were collected pre-and postoperatively. Perioperative treatments, neurological deficits, and rehabilitation therapies were recorded. Participants: The Neurospine group screened adult patients with intramedullary lesions scheduled for surgical intervention. The exclusion criteria included prior spinal surgery, neurological disorders, surgical complications, or relevant medications. Exposures The patients underwent standard intramedullary microsurgical procedures, received medications, and underwent assessments. The CSF protein profiles were examined using tandem mass spectrometry. Main outcomes and measures The evaluation of spinal function recovery involved various assessments and correlation analysis was performed. In addition, CSF protein features were identified and their clinical associations were assessed using least absolute shrinkage and selection operator (LASSO) regression modeling. Results Of the 43 patients who completed follow-up assessments, 23 were women and 20 were men, with a mean age of 42.8 ± 11.8 years. Histological diagnoses confirmed invasive lesions in 20 patients. Twenty-three patients demonstrated stable or improved spinal function, while 20 patients experienced deterioration during the 3rd and 6th month followups. As indicated by the changes in EMS scales (P = 0.02) and JOA scores (P = 0.03), functional prognosis was associated with pathological characteristics. We developed four models using the CSF proteomics datasets of noninvasive and invasive patients. These LASSO regression models exhibited high accuracy in predicting the recovery rates of the corresponding groups (P > 0.99). Conclusions and relevance This research demonstrates that lesion sizes and pathological characteristics are significant factors influencing spinal function recovery. The LASSO regression models developed using identified CSF protein variables for invasive and noninvasive patients, respectively, offer novel methods for the prediction of prognosis and the development of therapeutic approaches.

show abstract

“…Hereinafter, we briefly review some related works on omics data analysis. First of all, massive omics data often contain too many covariates but with only a few samples, considerable actually uncorrelated or independent covariates greatly hinder the subsequent analysis and applications ( Wang et al, 2022 ). Therefore, it is necessary to perform dimensional reduction or variable pre-filtering.…”

Section: Introductionmentioning

confidence: 99%

Exploring drought-responsive crucial genes in Sorghum

Wang

2022

iScience

View full text Add to dashboard Cite

Recent Advances on Penalized Regression Models for Biological Data

Cited by 5 publications

References 134 publications

The Effects of Artificial Diets on the Expression of Molecular Marker Genes Related to Honey Bee Health

The Effects of Artificial Diets on the Expression of Molecular Marker Genes Related to Honey Bee Health

Identification of Cerebrospinal Fluid Proteomic Signatures Associated with Intramedullary Surgery Outcome

Exploring drought-responsive crucial genes in Sorghum

Contact Info

Product

Resources

About