Skip to content

Latest commit

 

History

History
286 lines (203 loc) · 7.67 KB

CHANGELOG.md

File metadata and controls

286 lines (203 loc) · 7.67 KB
Raw

Changelog

0.26.0

  • Subsample the ecDNA distribution at the end of the simulation

0.25.0

  • Save based on the number of cells not time

0.24.0

  • Increase the number of max iterations and cells
  • Save at the end of the simulation
  • Specify either years or cells to simulate in the app

0.23.0

Huge update to remove useless stuff and keep only the content relevant for the PhD thesis. Remove abc in rust, do it in python.

0.22.0

change the way we save the data at the end of the simulations to simplify subsampling.

0.21.0

Assign idx such that it corresponds to the seed used.

0.20.1

Fix saving names: replace . by dot in the birth-death case.

0.20.0

change files and dir names after saving the data. Now: {SAMPLE}samples{CELLS}population/{MEASUREMENT}/{B0}b0_{B1}b1_{D0}d0_{D1}d1_0idx.json.

0.19.0

remove bin abc.

0.18.1

Save the absolute gillespie time at the last iteration. Takes 300 timepoints, that is save the dynamics max until time is equal to 30.

0.18.0

BugFix

Fix the dynamics as plotting a subset of them over time by binning gillespie time using fixed sized bins of time for all simulations

0.17.4

Added

Process PureBirthMeanTime.

0.17.3

Added

Process BirthDeathMeanTimeVarianceEntropy.

0.17.2

ecdna_lib version: 2.0.6

0.17.1

ecdna_lib version: 2.0.5

0.17.0

ecdna_lib version: 2.0.4

0.16.2

Fixed

  • Subsampling in the dynamics was incorrect, we now update the state after subsampling.
  • Take a reference of the process and the state instead of cloning it in the dynamics

Added

Subsampling for BirthDeathNMinusNPlusTime.

0.16.1

Fixed

Do not sample in BirthDeath process if there are not enough cells

0.16.0

ecdna_lib version: 2.0.0

0.15.0

Added

New flag drop_nminus for abc and ABCResult::run takes drop_nminus bool arg.

0.14.1

Added

ABC with cell-death assuming same rates for both cells with and without ecDNAs.

0.14.0

Fixed

  • saving twice the dynamics in some processes
  • not storing the NPlus and NMinus cells in processes BirthDeathMeanTimeVariance and BirthDeathMeanTime

Added

Remove restarting when cell-death, see issue #104.

0.13.12

Update readme.

0.13.11

Update readme.

0.13.10

ecdna_lib version: 1.0.0 ssa version: 1.0.0

0.13.9

The pure-birth process with time, nplus and nminus cells.

0.13.8

Fixed

Fix the birth-death process with time, nplus and nminus cells.

0.13.7

The birth-death process with time, nplus and nminus cells.

0.13.6

Fixed

Saving times only once.

0.13.5

The birth-death process with time, mean and variance.

ecdna_lib version: 0.5.4

0.13.4

The birth-death process with time and mean.

0.13.3

Remove simulate.

ecdna_lib version: 0.5.3

ssa version: 0.6.0

0.13.2

Using git tag attribute in Cargo.toml.

0.13.1

ecdna_lib version: 0.3.2

When subsample arg of clap is 0, take the full ecDNA distribution as a sample, which makes sense only if --sampling-strategy is not uniform.

0.13.0

ecdna_lib version: 0.3.0

New clap arg sampling_strategy and update ecdna-lib version to sample the ecDNA distributions according to different sampling strategies.

0.12.0

ssa version: 0.5.0

Refactor: rename PureBirthNoDynamics to PureBirth and PureBirth to PureBirthNMinusNPlus. The default process simulated by the dynamics is now PureBirth not PureBirthNoDynamics.

New flag nplus-nminus to generate PureBirthNMinusNPlus.

0.11.0

ssa version: 0.4.0

0.10.1

Update ecdna-evo version in Cargo.toml

0.10.0

This crate has now also its library, since in ssa v0.3.0 enum_dispatched heterogenous collection of processes has been removed.

ssa version: 0.3.0

0.9.0

Save a single abc.csv file for all runs instead of the abc folder with inside all runs.

0.8.1

Update ssa version

ssa version: 0.2.1

0.8.0

There are two main changes:

  1. refactor: create a new lib ecdna-lib and moved abc.rs and the ecDNA distribution there.
  2. Use ChaCha8 as rng instead of Pcg, so we can set the stream in independent threads, see here.

ssa version: 0.2.0

0.7.2

Update ssa version.

ssa version: 0.1.2

0.7.1

Some many bug fixes. The previous version (v.0.7.0) contains so much bugs that it should not be used.

ssa version: 0.1.1

0.7.0

The code uses now the ssa lib.

Perform ABC with subsampling (one timepoint only for now) with only pure-birth. ABC infers only the fitness coefficient, that is the birth-rate of cells with ecDNAs b1.

Advantages:

  • reduced by approximatively one half the code base
  • more general, should be directly applicable the other problems/processes (see ssa::hsc)

0.4.4

Fixed

  • Fix path to save data with cell death
  • Fix computation of the variance

Changes

  • Save ecDNA statistics also before subsampling dynamics
  • Add segregation mode to perform the nonminus experiment

0.4.3

Changes

  • Ignore the test data::tests::ecdna_undersample_reproducible_different_trials.
  • Add new dynamics Uneven which tracks the number of complete uneven segregation events.
  • Simplify any_individual_left by removing overflow checks.

0.4.2

Fixed

  • Sampling was performed with replacement, now without replacement (might need more memory)
  • Remove results folder from preprocess app
  • The computation of the ks_distance has changed by taking into account the loss of precision in converting u16 to f16
  • Dont panic when computing the variance when there are no N+ cells, but return 0

0.4.1

Remove data from the project: folder results is now in ~/ecdna-results.

Added

  • Option --savedir for preprocess

Fixed

  • Avoid sample twice when multiple timepoints before saving the statistics ( both abc and simulate).
  • Bug in the cell culture experiment with subsampling: restarting from one subsample from the whole distribution with correct idx, issue #65

0.4.0

Removed plotting library into another git folder

0.3.3

Added

  • When plotting abc results, the flag --export can save the runs that been filtered based on the thresholds (to for instance make the plots in Mathematica)

Fixed

  • Remove the CONFIG options from preprocess subcommand.

0.3.0

  • Merged preprocess into ecdna whihc has now three commands: simulate, abc and preprocess.

Added

  • New file timepoint in output of abc inference abc.tar.gz.

0.2.0

Added

  • Subsample option.

  • New binary preprocess.

  • Inference of multiple timepoints (longitudinal analysis) with patient and cell lines.

  • The patient growth strategy implies that after the first timepoint, tumour growth is restarted from the whole population. Whereas cell line growth strategy implies that the tumour growth restart from the subsample.

  • New python plots.

  • Codecov

Fixed

  • Tarballs with relative paths

0.1.1

Added

  • The ecDNA distribution is a vector of u16 The ecDNA distribution is a vector where each entries is a cell and the content is the number of ecDNA copies the cell has.

  • The ecDNA distribution is saved as a JSON, i.e. as a mapping where the keys are the ecDNA copies and the entries are the number of cells for each key (histogram).

  • Implement the state pattern to pass the ecDNA distribution by consuming the Run instead of deep copy.

  • Introduce subsampling: run the simulations for a number of cells, then run ABC with 100 subsamples for each run. In the end, the number of tested simulations will be 100 * the number of runs.

Fixed

  • Fix bug in the computation of the ks distance.