The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.

ASmiR R package

R interface to the supplied ASmiR trained models for classifying plant miRNA sequences as associated or not associated with cold, drought, heat, or salt stress.

Validation note: The package bundles the supplied ASmiR model objects and feature rankings and wraps the original PseKNC workflow. Validate predictions against suitable reference results before using them for scientific conclusions.

Requirements

• R >= 4.1
• Biostrings (Bioconductor), e1071 (CRAN)
• Python 3 available as python on PATH. You can set ASMiR_PYTHON to the full path of your Python executable.

install.packages(c("remotes", "e1071"))
if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")
BiocManager::install("Biostrings")
remotes::install_local("C:/path/to/ASmiR", dependencies = TRUE)

Predict with the bundled example

library(ASmiR)
fasta <- system.file("extdata", "examples", "gene_mirna.fasta", package = "ASmiR")
asmir_validate_sequences(fasta)
features <- asmir_features(fasta)
dim(features)  # expected 10 x 1379 for bundled example

cold <- asmir_predict(fasta, stress = "cold")
print(cold)

all_stresses <- asmir_predict(fasta, stress = "all")
write.csv(all_stresses, "ASmiR_predictions.csv", row.names = FALSE)

Workflow fidelity

For each stress, the package generates five PseKNC feature blocks with k=1..5 and parameters lambda=3 and w=0.2, combines them into 1,379 columns, selects the first N indices in the corresponding SVM-RFE file, and calls the matching bundled SVM model. N is cold=380, drought=272, heat=174, salt=340. RNA U is converted to DNA T only for the original PseKNC feature-generation step. Input validation accepts A/U/G/C only.

Output

Name, Stress, p_stress, Probability, and Label. p_stress is the raw probability for SVM class Y; Probability reproduces the original workflow’s confidence-like score max(p_stress, 1-p_stress). Label thresholds are 0.5 by default.

Version 0.1.1 note

The SVM prediction step calls e1071’s predict.svm method directly via getFromNamespace() to avoid the S3 dispatch error encountered with the bundled serialized SVM models. Restart R before reinstalling this version. Full prediction parity with the original ASmiR workflow should still be verified against known reference predictions.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.