Methods of identifying cellular replication timing signatures and methods of use thereof
AssigneeThe Florida State University Research Foundation, Inc.
InventorsDavid M. Gilbert, Juan Carlos Rivera-Mulia
Life of this patent
Estimating the term…
Abstract
Methods for identifying and classifying differences between biological samples are based on replication timing (RT) data. By comparing RT data for a test sample(s) to RT data for already characterized samples, one can identify differences and profile any new cell type or disease. These new methods allow for the detection of all the changes between distinct samples, many of which would escape detection by previous methods that discard any features showing any intra-sample variation.
Claims11
A method of identifying a cell type or cell types in a test sample of cells or plurality of test samples of cells by identifying a replication timing (RT) signature for the test sample of cells or plurality of test samples of cells, the method comprising:
- a) generating a RT profile for the test sample or RT profiles for the plurality of test samples wherein each RT profile is a log 2 ratio of signals from early and late S-phase fractions calculated in non-overlapping windows of 50-200 kb;
- b) generating RT profiles for a plurality of reference samples of distinct known cell types, wherein each RT profile is a log 2 ratio of signals from early and late S-phase fractions calculated in non-overlapping windows of 50-200 kb;
- c) removing pericentromeric and subtelomeric regions to remove large repetitive DNA regions from the test sample and reference sample RT profiles;
- (d) removing chromosome segments having invariant RT between the test and reference samples resulting in a collection of remaining chromosome segments that include all variation in RT between the test and reference samples, wherein the remaining chromosome segments are variant regions;
- (e) performing k-means clustering of the variant regions from step (d) and identifying groups of genomic regions that are non-overlapping windows of 50-200 kb with correlated RT patterns across all test and reference samples;
- (f) performing unsupervised hierarchical clustering of the test and reference samples from step (e) to construct a dendrogram that groups the test and reference samples in distinct branches, and identifying RT signatures for the test sample or plurality of test samples, wherein each RT signature is a k-means cluster identified in step (e) with an RT pattern specific for the dendrogram branch of the hierarchical clustering that groups the test sample or plurality of test samples;
- and
- (g) identifying the cell type of the test sample or plurality of test samples by establishing to which hierarchical clustering branch the test sample or plurality of test samples belong, wherein clustering of the test sample or plurality of test samples within the same branch of reference samples of a specific known cell type indicates that the identity of the test sample or plurality of test samples is the specific known cell type.
Dependencies are inferred from the claim text.