Table 1 Comparison of recommendations by transformer and RNN for multiple scientific analyses
From: Transformer-based tool recommendation system in Galaxy
Scientific analysis | Tool/Tool sequences | Ground truth | Transformer | RNN |
|---|---|---|---|---|
1. Variant analysis | snpeff_sars_cov_2 [36] | snpSift_filter, snpSift_extractFields, lofreq_filter, vcf2tsv, multiqc, collapse_dataset, CONVERTER_vcf_to_vcf_bgzip_0 [36] | multiqc, collapse_dataset, CONVERTER_vcf_to_vcf_bgzip_0, snpSift_extractFields, vcf2tsv, snpSift_filter, lofreq_filter, freebayes, mimodd_varcall, snpfreqplot, gemini_load, vcfcombine [35, 38] | multiqc, collapse_dataset, vcf2tsv, CONVERTER_vcf_to_vcf_bgzip_0, lofreq_filter, snpSift_extractFields, snpSift_filter, tb_variant_filter, mimodd_map, vcffilter2 [37, 40] |
2. Single-cell | anndata_import [52] | scanpy_filter, anndata_inspect, anndata_manipulate, ucsc_cell_browser, scanpy_inspect, scanpy_filter_cells [52, 53] | scanpy_filter_cells, anndata_inspect, ucsc_cell_browser, scanpy_filter, scanpy_inspect, anndata_manipulate, scanpy_normalise_data, scanpy_plot, anndata_ops, scanpy_remove_confounders, scanpy_integrate_harmony, scanpy_normalize, scpred_get_feature_space, scanpy_find_variable_genes, scpred_predict_labels, scpred_eigen_decompose [53, 54] | anndata_manipulate, scanpy_filter, scanpy_filter_cells, ucsc_cell_browser, scanpy_inspect, anndata_inspect, scanpy_plot, scanpy_normalise_data, scmap_scmap_cluster, scmap_scmap_cell, scanpy_filter_genes [52, 53, 55] |
3. Deep learning | keras_train_and_eval [56] | model_prediction, ml_visualization_ex [56] | ml_visualization_ex, model_prediction, plotly_regression-_performanc_plot, sklearn_discriminant_classifier, plotly_ml_performance_plots [56] | model_prediction, ml_visualization_ex [56], nn_classifier [56] |
4. Prote-omics | mass_spectrometry-_imaging_filtering, cardinal-_preprocessing, cardinal-_segmentations [41] | Filter1 | Filter1, cardinal_spectra_plots, cardinal_combine, cardinal_mz_images, cardinal_classification, cardinal_data_exporter, cardinal_quality_report, maldi_quant_preprocessing, cardinal_preprocessing, cardinal_segmentations, cardinal_filtering, mass_spectrometry-_imaging_filtering, maldi_quant_peak_detection [41] | Filter1 |