Exercise:AllergenicityPrediction ans - Revision history

Carol: /* PART C: Can you beat the baseline? */

2026-09-29T20:57:32Z

PART C: Can you beat the baseline?

@@ Line 227: / Line 227: @@
 I therefore modified the Random Forest to give greater weight to the minority allergen class:
-<syntaxhighlight lang="python">
 rf_classifier = RandomForestClassifier(
      n_estimators=100,
      class_weight="balanced",
      random_state=42
 The hypothesis is that increasing the importance of allergens during training may reduce the number of false negatives and improve sensitivity.
 === Conclusion ===

Carol: /* PART B: Can you beat BLAST? */

2026-09-29T20:55:31Z

PART B: Can you beat BLAST?

← Older revision		Revision as of 22:55, 29 September 2026
Line 151:		Line 151:
	For the example model:		For the example model:

	ROC-AUC ≈ 0.~~835~~		ROC-AUC ≈ 0.559

	Thus, although the model has poor recall at the default classification threshold, its predicted probabilities contain useful information for separating the two classes.		Thus, although the model has poor recall at the default classification threshold, its predicted probabilities contain useful information for separating the two classes.

Carol: /* PART B: Can you beat BLAST? */

2026-09-29T20:53:45Z

PART B: Can you beat BLAST?

← Older revision		Revision as of 22:53, 29 September 2026
Line 85:		Line 85:
	\|-		\|-
	\| Accuracy		\| Accuracy
	\| 0.~~834~~		\| 0.835
	\|-		\|-
	\| Precision		\| Precision
	\| 0.~~642~~		\| 0.616
	\|-		\|-
	\| Recall (Sensitivity)		\| Recall (Sensitivity)
	\| 0.~~141~~		\| 0.137
	\|-		\|-
	\| Specificity		\| Specificity
	\| 0.~~983~~		\| 0.982
	\|-		\|-
	\| MCC		\| MCC
	\| 0.~~245~~		\| 0.234
	\|-		\|-
	\| ROC-AUC		\| ROC-AUC
	\| 0.~~835~~		\| 0.559
	\|}		\|}

Line 208:		Line 208:

	The important conclusion is therefore not that "more trees are always better", but that enough trees are needed for stable predictions. Beyond this point, adding trees mainly increases computation time.		The important conclusion is therefore not that "more trees are always better", but that enough trees are needed for stable predictions. Beyond this point, adding trees mainly increases computation time.


	=== PART C: Can you beat the baseline? ===		=== PART C: Can you beat the baseline? ===

Carol at 20:51, 29 September 2026

2026-09-29T20:51:25Z

← Older revision		Revision as of 22:51, 29 September 2026
Line 20:		Line 20:
	A simple starting point is therefore to compare a protein against known allergens using sequence similarity.		A simple starting point is therefore to compare a protein against known allergens using sequence similarity.

	~~----~~

	*'''Q2: Is there any other tool beyond NetAllergen that can be used as a proxy to predict allergenicity in silico?'''		*'''Q2: Is there any other tool beyond NetAllergen that can be used as a proxy to predict allergenicity in silico?'''
Line 34:		Line 33:
	Thus, sequence similarity, protein-family membership and structural similarity can all be used as proxies for potential allergenicity, but experimental evidence is required to demonstrate an allergic response.		Thus, sequence similarity, protein-family membership and structural similarity can all be used as proxies for potential allergenicity, but experimental evidence is required to demonstrate an allergic response.

	~~----~~

	*'''Q3: Could immunotherapy potentially reduce allergic responses to proteins found in brown algae?'''		*'''Q3: Could immunotherapy potentially reduce allergic responses to proteins found in brown algae?'''
Line 75:		Line 73:

	The dataset therefore contains substantially more non-allergens than allergens.		The dataset therefore contains substantially more non-allergens than allergens.

	~~----~~

	*'''Q5: How well does it work?'''		*'''Q5: How well does it work?'''
Line 125:		Line 121:
	The high accuracy is therefore somewhat misleading. The model correctly identifies most non-allergens but misses a large fraction of the allergens.		The high accuracy is therefore somewhat misleading. The model correctly identifies most non-allergens but misses a large fraction of the allergens.

	~~----~~

	*'''Q6: Where does the model make mistakes?'''		*'''Q6: Where does the model make mistakes?'''
Line 145:		Line 140:
	The class imbalance in the dataset (3206 non-allergens versus 688 allergens) may contribute to this behaviour.		The class imbalance in the dataset (3206 non-allergens versus 688 allergens) may contribute to this behaviour.

	~~----~~

	*'''Q7: Why can AUC be more informative than accuracy?'''		*'''Q7: Why can AUC be more informative than accuracy?'''
Line 168:		Line 162:
	Importantly, ROC-AUC should be calculated using the predicted '''probabilities''' rather than the final binary predictions.		Importantly, ROC-AUC should be calculated using the predicted '''probabilities''' rather than the final binary predictions.

	~~----~~

	*'''Q8: How does the number of estimators (10, 100, 1000) affect model performance?'''		*'''Q8: How does the number of estimators (10, 100, 1000) affect model performance?'''
Line 219:		Line 212:
	=== PART C: Can you beat the baseline? ===		=== PART C: Can you beat the baseline? ===

	There is '''no single correct answer''' to this section~~. Students are expected to develop and evaluate their own modification~~.		There is '''no single correct answer''' to this section.

	One possible solution is shown below.		One possible solution is shown below.

	~~----~~

	*'''Q9: Provide a rationale for the change and report the new performance.'''		*'''Q9: Provide a rationale for the change and report the new performance.'''

Carol at 20:48, 29 September 2026

2026-09-29T20:48:57Z

Show changes

Carol: /* PART B: Can you beat BLAST */

2026-09-29T20:11:51Z

PART B: Can you beat BLAST

← Older revision		Revision as of 22:11, 29 September 2026
Line 11:		Line 11:
	'''Answer:'''		'''Answer:'''

	=== PART B: Can you beat BLAST ===		=== PART B: Can you beat BLAST? ===

Carol: /* Answers */

2026-09-29T20:11:41Z

Answers

← Older revision		Revision as of 22:11, 29 September 2026
Line 10:		Line 10:

	'''Answer:'''		'''Answer:'''

			=== PART B: Can you beat BLAST ===

Carol: /* Get the Data and Filter It */

2026-09-29T20:10:03Z

Get the Data and Filter It

← Older revision		Revision as of 22:10, 29 September 2026
Line 1:		Line 1:
	== Answers ==		== Answers ==

	=== ~~Get the Data and Filter It~~ ===		=== PART A: Investigating Allergenicity ===

	*'''Q1A:''' What would be your first approach to determine whether brown algae contain potential allergens? Find one database containing experimentally characterized allergens. Consider what biological information you would search for and what properties of the proteins might provide evidence of allergenic potential.		*'''Q1A:''' What would be your first approach to determine whether brown algae contain potential allergens? Find one database containing experimentally characterized allergens. Consider what biological information you would search for and what properties of the proteins might provide evidence of allergenic potential.

Carol: Created page with "== Answers == === Get the Data and Filter It === '''Q1A:''' What would be your first approach to determine whether brown algae contain potential allergens? Find one database containing experimentally characterized allergens. Consider what biological information you would search for and what properties of the proteins might provide evidence of allergenic potential. '''Answer:''' '''Q1B:''' Now imagine that no computational predictor specifically designed for allerg..."

2026-09-29T20:09:33Z

Created page with "== Answers == === Get the Data and Filter It === *'''Q1A:''' What would be your first approach to determine whether brown algae contain potential allergens? Find one database containing experimentally characterized allergens. Consider what biological information you would search for and what properties of the proteins might provide evidence of allergenic potential. '''Answer:''' *'''Q1B:''' Now imagine that no computational predictor specifically designed for allerg..."

New page

== Answers ==

=== Get the Data and Filter It ===

*'''Q1A:''' What would be your first approach to determine whether brown algae contain potential allergens? Find one database containing experimentally characterized allergens. Consider what biological information you would search for and what properties of the proteins might provide evidence of allergenic potential.

'''Answer:'''

*'''Q1B:''' Now imagine that no computational predictor specifically designed for allergenicity has yet been developed. How could you attempt to predict allergenicity in silico? Which existing protein predictions or sequence properties could potentially be used as proxies for allergenicity? There is not necessarily a single correct solution here. Think about what biological properties characterize allergens and what computational tools could provide information about those properties.

'''Answer:'''