Data Sets and Results for "Improved data sets and evaluation methods for the automatic prediction of DNA-binding proteins"

<p>Data sets and results for &quot;Improved data sets and evaluation methods for the automatic prediction of DNA-binding proteins&quot;<br> <br> The file &quot;dna_binding_protein_sequences.zip&quot; has the training and testing sets from the paper:<br> <br> RLL - &quot;random_&lt;train/test&gt;_full_1000.csv&quot;<br> RSL - &quot;random_&lt;train/test&gt;_40.csv&quot;<br> RS&amp;LL - &quot;random_&lt;train/test&gt;_40_1000.csv&quot;<br> RLL where included positive examples have verified DNA binding activity -&nbsp;&quot;random_&lt;train/test&gt;_hq_1000.csv&quot;<br> The 10 RS&amp;LL data sets - &quot;random_&lt;train/test&gt;_40_1000.csv&quot; +&nbsp;&quot;random_&lt;train/test&gt;_40_1000_cv_&lt;0-8&gt;.csv&quot;</p> <p>The results files are&nbsp;named similarly. See &quot;see_results.ipynb&quot; in the <a href="https://doi.org/10.5281/zenodo.5153683">codebase</a> that supplement these&nbsp;data sets<br> The species data sets are derived from &quot;uniprot_data_bac.tab&quot; and &quot;uniprot_data_not_bac.tab.&quot;&nbsp; See <a href="https://doi.org/10.5281/zenodo.5153683">code</a>.&nbsp;<br> <br> The ESM embeddings used by the XGBoost model are in &quot;dna_binding_protein_esm.zip&quot;</p>