Random forest algorithm reveals novel sites in HA protein that shift receptor binding preference of the H9N2 avian influenza virus
Bibliographic record
Abstract
After filtering out redundant, incomplete, environmental source, and unclear information sequences, a total of 5,656 H9N2 HA gene sequences were obtained from the National Center for Biotechnology Information (NCBI), Global Initiative on Sharing All Influenza Data (GISAID), and Influenza Research Database (IRD) databases. HA sequences were divided into two datasets based on host information: avian-derived sequences (5,588) and mammal-derived sequences (68) and labeled accordingly. Amino acid types were replaced with numerical values to transform the amino acid sequences into machine-readable vectors (Supplementary Table S1). We used the Random Forest Classifier method in Sklearn (1.0.22) to train the random forest classifier (Abraham et al., 2014), and in consideration of the large difference in the numbers of avian and non-avian sequences, we selected a balanced number of samples for training. The specific training parameters were: random _state=0，n _estimators=1500，oob _score=True，n _jobs=-1，class _weight='balanced'. During the model performance evaluation process, five-fold cross-validation was used to examine the classification performance of the random forest model, with the area under the ROC curve (AUC) used as the evaluation metric. Random under-sampling was applied to the avian-derived data during the training process. In the feature selection process, all data were used to train the random forest classifier and perform feature selection. After training, we extracted weight information for each site to represent its importance.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".