Machine learning enabled pulse and base calling for sequencing devices
A method includes obtaining, from one or more sequencing devices, raw data detected from luminescent labels associated with nucleotides during nucleotide incorporation events; and processing the raw data to perform a comparison of base calls produced by a learning enabled, automatic base calling module of the one or more sequencing devices with actual values associated with the raw data, wherein the base calls identify one or more individual nucleotides from the raw data. Based on the comparison, an update to the learning enabled, automatic base calling module is created using at least some of the obtained raw data, and the update is made available to the one or more sequencing devices.
1. A method for identifying nucleotides of a nucleic acid molecular composition about a sample, the method comprising:
using at least one computer hardware processor to perform:
accessing data obtained from detected light emissions by the sample after excitation, wherein the light emissions are responsive to a series of excitation light pulses and the data includes numbers of photons detected after each of at least some of the light pulses;
organizing the data as a data structure, wherein the data structure is a matrix or image, wherein each value in the matrix is representative of a number of photons detected within at least one time interval after at least some of the light pulses, and wherein each pixel of the image is representative of a number of photons detected within one of the intervals after one of the at least some of the light pulses;
providing the data structure as input to a trained deep learning model, wherein the deep learning model is trained to:
(a) extract one or more features from the data structure, and
(b) based on the one or more features, determine at least one probability value for a possible class, the possible class comprising data related to molecular composition of the sample, for each feature; and
receiving, from the deep learning model, data related to molecular composition of the sample.
2. The method of claim 1 , wherein the deep learning model comprises a convolutional neural network.
3. The method of claim 1 , wherein the deep learning model comprises a connectionist temporal classification (CTC)-fitted neural network model.
4. The method of claim 1 , wherein the data identifies nucleotides of a nucleic acid within the sample and comprises, for each of a plurality of nucleotides, a respective time series of values indicating probabilities that the nucleotide was incorporated into the nucleic acid.
5. The method of claim 4 , further comprising:
wherein the output identifying nucleotides of the nucleic acid comprises, for each of a plurality of nucleotides, a probability that the nucleotide was incorporated into the nucleic acid; and
the method further comprises identifying a first one of the plurality of nucleotides in the nucleic acid when the probability that the first nucleotide was incorporated into the nucleic acid exceeds a threshold probability.
6. The method of claim 1 , wherein providing the data as input to the trained deep learning model comprises:
organizing the data into a plurality of time periods including a first time period and a second time period; and
providing data for each of the plurality of time periods as an input to the trained deep learning model to obtain a corresponding output indicating at least one nucleotide of the nucleic acid.
7. The method of claim 6 , wherein an output corresponding to a respective time period provided as input to the trained deep learning model indicates, for each of a plurality of nucleotides, a value indicating a probability that the nucleotide was incorporated into a nucleic acid in the sample during a time period.
8. The method of claim 7 , wherein providing the data as input to the trained deep learning model comprises:
identifying a plurality of portions of the data, each portion corresponding to a respective one of the nucleotide incorporation events; and
providing each of the plurality of portions of the data as an input to the trained deep learning model to obtain an output corresponding to the portion of the data.
9. The method of claim 8 , wherein the output corresponding to the portion of the data identifies a nucleotide that was incorporated into the nucleic acid.
10. The method of claim 1 , further comprising:
accessing training data obtained from detected light emissions by luminescent labels associated with nucleotides during nucleotide incorporation events for a plurality of nucleic acids of the sample; and
training a deep learning model using the training data and information specifying at least some of the nucleotides in the plurality of nucleic acids to obtain the trained deep learning model.
11. The method of claim 1 , wherein the data includes, for each of at least some of the light pulses, a respective number of photons detected in each of a plurality of intervals of a time period after the light pulse.
12. A system for identifying molecular information about a sample, the system comprising:
at least one computer hardware processor; and
at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform:
accessing data obtained from detected light emissions by the sample after excitation, wherein the light emissions are responsive to a series of excitation light pulses and the data includes numbers of photons detected after each of at least some of the light pulses;
organizing the data as a data structure, wherein the data structure is a matrix or image, wherein each value in the matrix is representative of a number of photons detected within at least one time interval after at least some of the light pulses, and wherein each pixel of the image is representative of a number of photons detected within one of the intervals after one of the at least some of the light pulses;
providing the data structure as input to a trained deep learning model, wherein the deep learning model is trained to:
(a) extract one or more features from the data structure, and
(b) based on the one or more features, determine at least one probability value for a possible class, the possible class comprising data related to molecular composition of the sample, for each feature; and
receiving, from the deep learning model, data identifying molecular information about the sample.
13. The system of claim 12 , wherein the data includes data indicating a respective number of photons in each of a first and a second time interval in the first time period after a first light pulse in the series of light pulses.
14. The system of claim 13 , wherein the data further includes a respective number of photons in each of the first and second time intervals in the second time period after a second light pulse in the series of light pulses.
15. The system of claim 14 , wherein providing the data as input to the trained machine learning model comprises arranging the data in an image, wherein each pixel of the image specifies a number of photons detected in an interval of a time period after one of the at least some pulses.
16. The system of claim 14 , wherein providing the data as input to the trained machine learning model comprises arranging the data into a data structure having rows wherein each of the rows holds numbers of photons in a respective interval corresponding to the at least some light pulses.
17. The system of claim 14 , wherein providing the data as input to the trained machine learning model comprises:
identifying a plurality of portions of the data, each portion corresponding to a respective one of the nucleotides of the nucleic acid; and
providing each of the plurality of portions of the data as an input to the trained machine learning model to obtain an output corresponding to the portion of the data.
18. The system of claim 14 , further comprising:
accessing training data obtained from detected light emissions by luminescent labels associated with nucleotides during nucleotide incorporation events for a plurality of nucleic acids of the sample; and
training a machine learning model using the training data and information specifying at least some of the nucleotides in the plurality of nucleic acids to obtain the trained machine learning model.
19. At least one non-transitory computer-readable storage medium storing instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform:
accessing data obtained from detected light emissions by a sample after excitation, wherein the light emissions are responsive to a series of excitation light pulses and the data includes numbers of photons detected after each of at least some of the light pulses;
organizing the data as a data structure, wherein the data structure is a matrix or image, wherein each value in the matrix is representative of a number of photons detected within at least one time interval after at least some of the light pulses, and wherein each pixel of the image is representative of a number of photons detected within one of the intervals after one of the at least some of the light pulses;
providing the data structure as input to a trained deep learning model, wherein the deep learning model is trained to:
(a) extract one or more features from the data structure, and
(b) based on the one or more features, determine at least one probability value for a possible class, the possible class comprising data related to molecular composition of the sample, for each feature; and
receiving, from the deep learning model, data identifying molecular information about the sample.