Predict Modality

Predicting the profiles of one modality (e.g. protein abundance) from another (e.g. mRNA expression).

Repository: openproblems-bio/task_predict_modality

Description

Experimental techniques to measure multiple modalities within the same single cell are increasingly becoming available. The demand for these measurements is driven by the promise to provide a deeper insight into the state of a cell. Yet, the modalities are also intrinsically linked. We know that DNA must be accessible (ATAC data) to produce mRNA (expression data), and mRNA in turn is used as a template to produce protein (protein abundance). These processes are regulated often by the same molecules that they produce: for example, a protein may bind DNA to prevent the production of more mRNA. Understanding these regulatory processes would be transformative for synthetic biology and drug target discovery. Any method that can predict a modality from another must have accounted for these regulatory processes, but the demand for multi-modal data shows that this is not trivial.

Authors & contributors

Name	Roles	Github	Orcid	Email
Alejandro Granados	author	agranado
Alex Tong	author	atong01
Bastian Rieck	author	Pseudomanifold
Christopher Lance	author	xlancelottx	0000-0002-1275-9802
Daniel Burkhardt	author	dburkhardt
Kai Waldrant	contributor	KaiWaldrant	0009-0003-8555-1361
Kaiwen Deng	contributor	nonztalk		dengkw@umich.edu
Louise Deconinck	author	LouiseDck
Robrecht Cannoodt	author, maintainer	rcannood	0000-0003-3641-729X
Xueer Chen	contributor	xuerchen		xc2579@columbia.edu
Jiwei Liu	contributor	daxiongshu	0000-0002-8799-9763	jiweil@nvidia.com
Marius Lange	contributor	marius1311	0000-0002-4846-1266

API

flowchart TB
  file_common_dataset_mod1("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-raw-dataset-rna'>Raw dataset RNA</a>")
  comp_process_datasets[/"<a href='https://github.com/openproblems-bio/task_predict_modality#component-type-process-dataset'>Process Dataset</a>"/]
  file_train_mod1("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-train-mod1'>Train mod1</a>")
  file_train_mod2("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-train-mod2'>Train mod2</a>")
  file_test_mod1("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-test-mod1'>Test mod1</a>")
  file_test_mod2("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-solution'>Solution</a>")
  comp_control_method[/"<a href='https://github.com/openproblems-bio/task_predict_modality#component-type-control-method'>Control method</a>"/]
  comp_method[/"<a href='https://github.com/openproblems-bio/task_predict_modality#component-type-method'>Method</a>"/]
  comp_method_predict[/"<a href='https://github.com/openproblems-bio/task_predict_modality#component-type-predict'>Predict</a>"/]
  comp_method_train[/"<a href='https://github.com/openproblems-bio/task_predict_modality#component-type-train'>Train</a>"/]
  comp_metric[/"<a href='https://github.com/openproblems-bio/task_predict_modality#component-type-metric'>Metric</a>"/]
  file_prediction("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-prediction'>Prediction</a>")
  file_pretrained_model("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-pretrained-model'>Pretrained model</a>")
  file_score("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-score'>Score</a>")
  file_common_dataset_mod2("<a href='https://github.com/openproblems-bio/task_predict_modality#file-format-raw-dataset-mod2'>Raw dataset mod2</a>")
  file_common_dataset_mod1---comp_process_datasets
  comp_process_datasets-->file_train_mod1
  comp_process_datasets-->file_train_mod2
  comp_process_datasets-->file_test_mod1
  comp_process_datasets-->file_test_mod2
  file_train_mod1---comp_control_method
  file_train_mod1---comp_method
  file_train_mod1-.-comp_method_predict
  file_train_mod1---comp_method_train
  file_train_mod2---comp_control_method
  file_train_mod2---comp_method
  file_train_mod2-.-comp_method_predict
  file_train_mod2---comp_method_train
  file_test_mod1---comp_control_method
  file_test_mod1---comp_method
  file_test_mod1---comp_method_predict
  file_test_mod1-.-comp_method_train
  file_test_mod2---comp_control_method
  file_test_mod2---comp_metric
  comp_control_method-->file_prediction
  comp_method-->file_prediction
  comp_method_predict-->file_prediction
  comp_method_train-->file_pretrained_model
  comp_metric-->file_score
  file_prediction---comp_metric
  file_pretrained_model---comp_method_predict
  file_common_dataset_mod2---comp_process_datasets

Loading

File format: Raw dataset RNA

The RNA modality of the raw dataset.

Example file: resources_test/common/openproblems_neurips2021/bmmc_cite/dataset_mod1.h5ad

Format:

AnnData object
 obs: 'batch', 'size_factors'
 var: 'feature_id', 'feature_name', 'hvg', 'hvg_score'
 obsm: 'gene_activity'
 layers: 'counts', 'normalized'
 uns: 'dataset_id', 'dataset_name', 'dataset_url', 'dataset_reference', 'dataset_summary', 'dataset_description', 'dataset_organism', 'normalization_id', 'gene_activity_var_names'

Data structure:

Slot	Type	Description
obs["batch"]	string	Batch information.
obs["size_factors"]	double	(Optional) The size factors of the cells prior to normalization.
var["feature_id"]	string	Unique identifier for the feature, usually a ENSEMBL gene id.
var["feature_name"]	string	(Optional) A human-readable name for the feature, usually a gene symbol.
var["hvg"]	boolean	Whether or not the feature is considered to be a ‘highly variable gene’.
var["hvg_score"]	double	A score for the feature indicating how highly variable it is.
obsm["gene_activity"]	double	(Optional) ATAC gene activity.
layers["counts"]	integer	Raw counts.
layers["normalized"]	double	Normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["dataset_name"]	string	Nicely formatted name.
uns["dataset_url"]	string	(Optional) Link to the original source of the dataset.
uns["dataset_reference"]	string	(Optional) Bibtex reference of the paper in which the dataset was published.
uns["dataset_summary"]	string	Short description of the dataset.
uns["dataset_description"]	string	Long description of the dataset.
uns["dataset_organism"]	string	(Optional) The organism of the sample in the dataset.
uns["normalization_id"]	string	The unique identifier of the normalization method used.
uns["gene_activity_var_names"]	string	(Optional) Names of the gene activity matrix.

Component type: Process Dataset

A predict modality dataset processor.

Arguments:

Name	Type	Description
--input_mod1	file	The RNA modality of the raw dataset.
--input_mod2	file	The second modality of the raw dataset. Must be an ADT or an ATAC dataset.
--output_train_mod1	file	(Output) The mod1 expression values of the train cells.
--output_train_mod2	file	(Output) The mod2 expression values of the train cells.
--output_test_mod1	file	(Output) The mod1 expression values of the test cells.
--output_test_mod2	file	(Output) The ground-truth mod2 expression values of the test cells.

File format: Train mod1

The mod1 expression values of the train cells.

Example file: resources_test/task_predict_modality/openproblems_neurips2021/bmmc_cite/swap/train_mod1.h5ad

Format:

AnnData object
 obs: 'batch', 'size_factors'
 var: 'gene_ids', 'hvg', 'hvg_score'
 obsm: 'gene_activity'
 layers: 'counts', 'normalized'
 uns: 'dataset_id', 'common_dataset_id', 'modality', 'dataset_organism', 'normalization_id', 'gene_activity_var_names'

Data structure:

Slot	Type	Description
obs["batch"]	string	Batch information.
obs["size_factors"]	double	(Optional) The size factors of the cells prior to normalization.
var["gene_ids"]	string	(Optional) The gene identifiers (if available).
var["hvg"]	boolean	Whether or not the feature is considered to be a ‘highly variable gene’.
var["hvg_score"]	double	A score for the feature indicating how highly variable it is.
obsm["gene_activity"]	double	(Optional) ATAC gene activity.
layers["counts"]	integer	Raw counts.
layers["normalized"]	double	Normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["common_dataset_id"]	string	(Optional) A common identifier for the dataset.
uns["modality"]	string	The modality of the data, e.g. ‘GEX’, ‘ADT’ or ‘ATAC’. Methods use this to pick modality-specific hyperparameters or models.
uns["dataset_organism"]	string	(Optional) The organism of the sample in the dataset.
uns["normalization_id"]	string	The unique identifier of the normalization method used.
uns["gene_activity_var_names"]	string	(Optional) Names of the gene activity matrix.

File format: Train mod2

The mod2 expression values of the train cells.

Example file: resources_test/task_predict_modality/openproblems_neurips2021/bmmc_cite/swap/train_mod2.h5ad

Format:

AnnData object
 obs: 'batch', 'size_factors'
 var: 'gene_ids', 'hvg', 'hvg_score'
 obsm: 'gene_activity'
 layers: 'counts', 'normalized'
 uns: 'dataset_id', 'common_dataset_id', 'modality', 'dataset_organism', 'normalization_id', 'gene_activity_var_names'

Data structure:

Slot	Type	Description
obs["batch"]	string	Batch information.
obs["size_factors"]	double	(Optional) The size factors of the cells prior to normalization.
var["gene_ids"]	string	(Optional) The gene identifiers (if available).
var["hvg"]	boolean	Whether or not the feature is considered to be a ‘highly variable gene’.
var["hvg_score"]	double	A score for the feature indicating how highly variable it is.
obsm["gene_activity"]	double	(Optional) ATAC gene activity.
layers["counts"]	integer	Raw counts.
layers["normalized"]	double	Normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["common_dataset_id"]	string	(Optional) A common identifier for the dataset.
uns["modality"]	string	The modality of the data, e.g. ‘GEX’, ‘ADT’ or ‘ATAC’. Methods use this to pick modality-specific hyperparameters or models.
uns["dataset_organism"]	string	(Optional) The organism of the sample in the dataset.
uns["normalization_id"]	string	The unique identifier of the normalization method used.
uns["gene_activity_var_names"]	string	(Optional) Names of the gene activity matrix.

File format: Test mod1

The mod1 expression values of the test cells.

Example file: resources_test/task_predict_modality/openproblems_neurips2021/bmmc_cite/swap/test_mod1.h5ad

Format:

AnnData object
 obs: 'batch', 'size_factors'
 var: 'gene_ids', 'hvg', 'hvg_score'
 obsm: 'gene_activity'
 layers: 'counts', 'normalized'
 uns: 'dataset_id', 'common_dataset_id', 'modality', 'dataset_name', 'dataset_url', 'dataset_reference', 'dataset_summary', 'dataset_description', 'dataset_organism', 'normalization_id', 'gene_activity_var_names'

Data structure:

Slot	Type	Description
obs["batch"]	string	Batch information.
obs["size_factors"]	double	(Optional) The size factors of the cells prior to normalization.
var["gene_ids"]	string	(Optional) The gene identifiers (if available).
var["hvg"]	boolean	Whether or not the feature is considered to be a ‘highly variable gene’.
var["hvg_score"]	double	A score for the feature indicating how highly variable it is.
obsm["gene_activity"]	double	(Optional) ATAC gene activity.
layers["counts"]	integer	Raw counts.
layers["normalized"]	double	Normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["common_dataset_id"]	string	(Optional) A common identifier for the dataset.
uns["modality"]	string	The modality of the data, e.g. ‘GEX’, ‘ADT’ or ‘ATAC’. Methods use this to pick modality-specific hyperparameters or models.
uns["dataset_name"]	string	Nicely formatted name.
uns["dataset_url"]	string	(Optional) Link to the original source of the dataset.
uns["dataset_reference"]	string	(Optional) Bibtex reference of the paper in which the dataset was published.
uns["dataset_summary"]	string	Short description of the dataset.
uns["dataset_description"]	string	Long description of the dataset.
uns["dataset_organism"]	string	(Optional) The organism of the sample in the dataset.
uns["normalization_id"]	string	The unique identifier of the normalization method used.
uns["gene_activity_var_names"]	string	(Optional) Names of the gene activity matrix.

File format: Solution

The ground-truth mod2 expression values of the test cells.

Example file: resources_test/task_predict_modality/openproblems_neurips2021/bmmc_cite/swap/test_mod2.h5ad

Description:

The ground truth against which predictions are scored. Only the metrics and the control methods receive this file; regular methods never see it.

Format:

AnnData object
 obs: 'batch', 'size_factors'
 var: 'gene_ids', 'hvg', 'hvg_score'
 obsm: 'gene_activity'
 layers: 'counts', 'normalized'
 uns: 'dataset_id', 'common_dataset_id', 'modality', 'dataset_name', 'dataset_url', 'dataset_reference', 'dataset_summary', 'dataset_description', 'dataset_organism', 'gene_activity_var_names'

Data structure:

Slot	Type	Description
obs["batch"]	string	Batch information.
obs["size_factors"]	double	(Optional) The size factors of the cells prior to normalization.
var["gene_ids"]	string	(Optional) The gene identifiers (if available).
var["hvg"]	boolean	Whether or not the feature is considered to be a ‘highly variable gene’.
var["hvg_score"]	double	A score for the feature indicating how highly variable it is.
obsm["gene_activity"]	double	(Optional) ATAC gene activity.
layers["counts"]	integer	Raw counts.
layers["normalized"]	double	Normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["common_dataset_id"]	string	(Optional) A common identifier for the dataset.
uns["modality"]	string	The modality of the data, e.g. ‘GEX’, ‘ADT’ or ‘ATAC’. Methods use this to pick modality-specific hyperparameters or models.
uns["dataset_name"]	string	Nicely formatted name.
uns["dataset_url"]	string	(Optional) Link to the original source of the dataset.
uns["dataset_reference"]	string	(Optional) Bibtex reference of the paper in which the dataset was published.
uns["dataset_summary"]	string	Short description of the dataset.
uns["dataset_description"]	string	Long description of the dataset.
uns["dataset_organism"]	string	(Optional) The organism of the sample in the dataset.
uns["gene_activity_var_names"]	string	(Optional) Names of the gene activity matrix.

Component type: Control method

Quality control methods for verifying the pipeline.

Arguments:

Name	Type	Description
--input_train_mod1	file	The mod1 expression values of the train cells.
--input_train_mod2	file	The mod2 expression values of the train cells.
--input_test_mod1	file	The mod1 expression values of the test cells.
--input_test_mod2	file	The ground-truth mod2 expression values of the test cells.
--output	file	(Output) A prediction of the mod2 expression values of the test cells.

Component type: Method

A regression method.

Arguments:

Name	Type	Description
--input_train_mod1	file	The mod1 expression values of the train cells.
--input_train_mod2	file	The mod2 expression values of the train cells.
--input_test_mod1	file	The mod1 expression values of the test cells.
--output	file	(Output) A prediction of the mod2 expression values of the test cells.

Component type: Predict

Make predictions using a trained model.

Arguments:

Name	Type	Description
--input_train_mod1	file	(Optional) The mod1 expression values of the train cells.
--input_train_mod2	file	(Optional) The mod2 expression values of the train cells.
--input_test_mod1	file	The mod1 expression values of the test cells.
--input_model	file	A pretrained model for predicting the expression of one modality from another.
--output	file	(Output) A prediction of the mod2 expression values of the test cells.

Component type: Train

Train a model to predict the expression of one modality from another.

Arguments:

Name	Type	Description
--input_train_mod1	file	The mod1 expression values of the train cells.
--input_train_mod2	file	The mod2 expression values of the train cells.
--input_test_mod1	file	(Optional) The mod1 expression values of the test cells.
--output	file	(Output) A pretrained model for predicting the expression of one modality from another.

Component type: Metric

A predict modality metric.

Arguments:

Name	Type	Description
--input_prediction	file	A prediction of the mod2 expression values of the test cells.
--input_test_mod2	file	The ground-truth mod2 expression values of the test cells.
--output	file	(Output) Metric score file.

File format: Prediction

A prediction of the mod2 expression values of the test cells

Example file: resources_test/task_predict_modality/openproblems_neurips2021/bmmc_cite/swap/prediction.h5ad

Format:

AnnData object
 layers: 'normalized'
 uns: 'dataset_id', 'method_id'

Data structure:

Slot	Type	Description
layers["normalized"]	double	Predicted normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["method_id"]	string	A unique identifier for the method.

File format: Pretrained model

A pretrained model for predicting the expression of one modality from another.

File format: Score

Metric score file

Example file: resources_test/task_predict_modality/openproblems_neurips2021/bmmc_cite/swap/score.h5ad

Format:

AnnData object
 uns: 'dataset_id', 'method_id', 'metric_ids', 'metric_values'

Data structure:

Slot	Type	Description
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["method_id"]	string	A unique identifier for the method.
uns["metric_ids"]	string	One or more unique metric identifiers.
uns["metric_values"]	double	The metric values obtained for the given prediction. Must be of same length as ‘metric_ids’.

File format: Raw dataset mod2

The second modality of the raw dataset. Must be an ADT or an ATAC dataset

Example file: resources_test/common/openproblems_neurips2021/bmmc_cite/dataset_mod2.h5ad

Format:

AnnData object
 obs: 'batch', 'size_factors'
 var: 'feature_id', 'feature_name', 'hvg', 'hvg_score'
 obsm: 'gene_activity'
 layers: 'counts', 'normalized'
 uns: 'dataset_id', 'dataset_name', 'dataset_url', 'dataset_reference', 'dataset_summary', 'dataset_description', 'dataset_organism', 'normalization_id', 'gene_activity_var_names'

Data structure:

Slot	Type	Description
obs["batch"]	string	Batch information.
obs["size_factors"]	double	(Optional) The size factors of the cells prior to normalization.
var["feature_id"]	string	Unique identifier for the feature, usually a ENSEMBL gene id.
var["feature_name"]	string	(Optional) A human-readable name for the feature, usually a gene symbol.
var["hvg"]	boolean	Whether or not the feature is considered to be a ‘highly variable gene’.
var["hvg_score"]	double	A score for the feature indicating how highly variable it is.
obsm["gene_activity"]	double	(Optional) ATAC gene activity.
layers["counts"]	integer	Raw counts.
layers["normalized"]	double	Normalized expression values.
uns["dataset_id"]	string	A unique identifier for the dataset.
uns["dataset_name"]	string	Nicely formatted name.
uns["dataset_url"]	string	(Optional) Link to the original source of the dataset.
uns["dataset_reference"]	string	(Optional) Bibtex reference of the paper in which the dataset was published.
uns["dataset_summary"]	string	Short description of the dataset.
uns["dataset_description"]	string	Long description of the dataset.
uns["dataset_organism"]	string	(Optional) The organism of the sample in the dataset.
uns["normalization_id"]	string	The unique identifier of the normalization method used.
uns["gene_activity_var_names"]	string	(Optional) Names of the gene activity matrix.