JPWO2020191389A5

JPWO2020191389A5 -

Info

Publication number: JPWO2020191389A5
Application number: JP2020572704A
Authority: JP
Publication date: 2023-03-30

Claims

A computer-implemented method for generating ground truth training data for training a neural network-based template generator for a cluster metadata determination task, comprising:
accessing a series of image sets generated during a sequencing run, each image set generated during a respective sequencing cycle of the sequencing run, each image set comprising: drawing clusters and their surrounding background, each image of the series comprising pixels in a pixel domain, each of the pixels being divided into a plurality of sub-pixels in a sub-pixel domain ;
obtaining from a base caller a base call that classifies each of said subpixels as one of four bases (A, C, T, and G), thereby over multiple sequencing cycles of said sequencing operation: generating a base call sequence for each of the sub-pixels;
generating a cluster map that identifies the clusters as discrete regions of adjacent sub-pixels that share substantially matching base call sequences;
determining cluster metadata based on the discrete regions in the cluster map;
determining that the cluster metadata includes cluster centers, cluster shapes, cluster sizes, cluster backgrounds, and/or cluster boundaries;
Generating ground truth training data for using the cluster metadata to train a neural network-based template generator for a cluster metadata determination task, comprising :
the ground truth training data comprises an attenuation map, a ternary map, or a binary map;
generating , wherein the neural network-based template generator is trained to generate the attenuation map, the ternary map, or the binary map as an output based on the ground truth training data;
computer-implemented methods, including

said attenuation map, said ternary map, or generated as output by a neural network-based template generator for base calling by a neural network-based base caller to increase throughput in high-throughput nucleic acid sequencing technology; 2. The computer-implemented method of claim 1 , further comprising using the cluster metadata derived from the binary map.

3. The computer-implemented method of claim 1 or 2, further comprising generating the cluster map by identifying sub-pixels that do not belong to any of the discrete regions as background.

4. The computer-implemented method of any one of claims 1-3, wherein the cluster map identifies cluster boundary portions between two consecutive sub-pixels where the base call sequences do not substantially match.

The cluster map is
identifying an origin sub-pixel at preliminary center coordinates of said cluster determined by a base caller;
5. The computer of any one of claims 1 to 4, wherein, starting from the origin subpixel, and continuing with successive non-origin subpixels, it performs a breadth-first search for substantially matching base-call sequences. How to implement .

determining hyper-location center coordinates of the cluster by calculating the center of mass of the discrete region of the cluster map as the average of the coordinates of each successive sub-pixel forming the discrete region;
storing the hyper-location center coordinates of the clusters in memory for use as the ground truth training data for training the neural network - based template generator. 10. The computer-implemented method of any one of Clause 1 .

identifying the center of a mass sub-pixel within a disjoint region of the cluster map at the hyper-located center coordinates of the cluster;
Upsample the cluster map using interpolation and store the upsampled cluster map in the memory for use as the ground truth training data for training the neural network-based template generator. and
In an upsampled cluster map, each contiguous sub-pixel within a discontinuous region to which it belongs is based on an attenuation coefficient proportional to the distance of the neighboring sub-pixel from the center of the mass sub-pixel within said discontinuous region. 7. The computer-implemented method of claim 6 , further comprising assigning values to pixels.

generating the attenuation map from the upsampled cluster map representing the contiguous sub-pixels within the discontinuous region, wherein the sub-pixels are identified as the background based on the assigned values;
8. The computer-implemented method of claim 7 , further comprising storing the attenuation map in the memory for use as the ground truth training data for training the neural network-based template generator. .

In the upsampled cluster map, for each cluster, classifying the contiguous sub-pixels within the non-junction region as cluster-inner sub-pixels belonging to the same cluster; classifying pixel centers, sub-pixels containing cluster boundary portions, and sub-pixels identified as background as background sub-pixels;
9. The computer-implemented method of claim 8 , further comprising storing the classification in the memory for use as the ground truth training data for training the neural network-based template generator.

memory for using, for each cluster, cluster interior subpixels, cluster center subpixels, boundary subpixels, and background subpixels as the ground truth training data for training the neural network-based template generator; storing the background subpixels in
downscaling coordinates by a factor used to upsample the cluster map;
and storing the downscaled coordinates in the memory for use as the ground truth training data for training the neural network-based template generator, for each cluster. 10. The computer-implemented method of any one of Claims 1-9 .

generating a cluster map of multiple tiles of the flow cell;
storing the cluster map in memory, and cluster metadata for clusters within the cluster based on the cluster map, including the cluster center, the cluster shape, the cluster size, the cluster background, and/or the cluster boundary; and
In an upsampled cluster map of said clusters within said tile, classifying sub-pixels by cluster by cluster and sub-pixels as cluster inside sub-pixels belonging to the same cluster, cluster center sub-pixels, boundary sub-pixels. , and background subpixels;
storing the classification in the memory for use as the ground truth training data for training the neural network-based template generator;
the coordinates of the cluster interior sub-pixels, the cluster center sub-pixels, the boundary sub-pixels, and the background sub-pixels, for each cluster in the cluster, the ground truth training for training the neural network-based template generator; storing the background subpixels in the memory for use as data;
downscaling the coordinates by a factor used to upsample the cluster map;
and storing the downscaled coordinates in the memory for use as the ground truth training data for training the neural network-based template generator for each cluster across the tiles. , a computer-implemented method according to any one of claims 1-10 .

12. The computer-implemented method of any one of claims 1-11, wherein the base call sequences substantially match when predetermined portions of the base calls match for each ordinal position.

13. The computer-implemented method of any one of claims 1-12, wherein the cluster map is generated based on a predetermined minimum number of sub-pixels for discontinuous regions.

the flow cell has at least one patterned surface with an array of wells occupying the clusters;
based on the determined shape and size of the clusters, determining one of the wells is substantially occupied by at least one cluster;
one of said wells is minimally occupied;
14. The computer-implemented method of any one of claims 1-13, wherein one of the wells is co-occupied by a plurality of populations.