CDBProm v2.0 — Prediction API Documentation

CDBProm v2.0 — Prediction API

This API provides programmatic access to the promoter prediction functionality used by CDBProm v2.0.
Users can submit their own sequences to the prediction API.

For using CDBProm's API, you will need to request an API key through our contact page

Request Format

Endpoint: https://xxx/predict/

Content-type: application/json

Request Body (JSON)

The request must be sent as a JSON object containing a list of sequences in FASTA format.

Parameter	Type	Required	Description
id	String	Yes	FASTA header of the sequence
seq	String	Yes	Sequence in nucleotides

Input Structure

                    "sequences": [
                        {
                            "id": "seq1",
                            "seq": "TTTCGTAACTGGGTCGGATTTCCTGCAAGATCCTTTGTGCCTCCGGAGGGGGCTCGTGCC"
                        }
                        {
                            "id": "seq2",
                            "seq": "GCCGGGACGCACGGGACCGTCGGCGCAAGTATGCGCAACCTGGTCACAGAGGAGAGCGTC"
                        }]

Constraints

Requests violating these constraints will return an error.

Constraint	Value
Maximum sequences per request	100
Minimum sequence length	60
Maximum sequence length	1000

For requests exceding 100 sequences, please refer to the local implementation with our Docker image.

Response Structure

Field	Description
id	FASTA header.
Coordinates	Coordinates on the sequence where the prediction was made (for sequences with length > 60, a subsequences of length 60 considering a step size of 10 are generated).
Probability non-promoter	Probability (0-100%) of a sequence beloning to the non-promoter class (output of the softmax layer)
Probability promoter	Probability (0-100%) of a sequence beloning to the promoter class (output of the softmax layer)
Predicted class	Predicted class of a sequence (≥0.5 are promoters)
Message	Message indicating how the API treated the input sequence Success Error. Sequence with length greater than 1000. The maximum length per sequence is 1000. Input sequence(s) too short. The minimum length is 60.

Response Structure

                    "output": [
                        {
                            "id": "seq1",
                            "Coordinates": "1-60",
                            "Predicted class": "Promoter",
                            "Probability promoter": "98.94%",
                            "Probability non-promoter": "1.06%",
                            "Sequence": "TTTCGTAACTGGGTCGGATTTCCTGCAAGATCCTTTGTGCCTCCGGAGGGGGCTCGTGCC"
                        }]