HMM-Flagger v1.1.0

Changes since v1.0.0:

• Tuned HMM-Flagger for different platforms
• Updated mobinasri/long_read_aligner:v1.1.0 with minimap2-2.28
• Added minHighMapqRatio parameter for restricting transition to collapsed state (with the default value of 0.75)
• Reduced the default value of mapqThreshold to 10 for bam2cov program
• Added a WDL parameter for decomposing censat bed file. With this option censat annotation bed file that is created with centromereAnnotation.wdl can be passed directly to HMM-Flagger WDLs.
• Added --writePosteriorProbs option (for development purposes)
• Added --startOnly mode to consider only the start locations of reads for computing coverage (experimental)

HMM-Flagger v1.0.0

Update README.md

Flagger v0.4.0

• WDLs are now taking both compressed and uncompressed fasta files
• Added a wdl for running mapping and flagger together (flagger_end_to_end_with_mapping.wdl)
• Removed long_read_aligner_secphase.wdl since long_read_aligner_scattered.wdl has now the option of running secphase
• Added parameter_meta
• flagger_end_to_end does not perform variant calling any more
• Tasks like running secphase, projecting blocks, detecting biases and adding MD tag are optional
• Added a task for automatically detecting annotations with coverage biases
• Reduced the margin size for merging biased blocks of the same category. 50kb leads to some FP collapsed blocks in monomeric regions beside hsat3
• Added instructions for running some test datasets on Slurm for long_read_aligner_scattered.wdl, flagger_end_to_end.wdl and flagger_end_to_end_with_mapping.wdl

Flagger v0.3.3

• Fixed a bug related to the size() function in WDL while running the pipeline locally with cromwell.
• Minor Update in README

Flagger v0.3.2

• Updated Secphase to 0.4.2 since in this version some major bugs in the marker mode are fixed and its runtime is improved noticeably
• Removed short unreliable blocks from the output

Flagger v0.3.1

Changes since v0.3.0

• Fixed a bug in split_cov_by_window to handle cov blocks longer than window size
• When cromwell is run locally it has an issue with the size() function. It cannot get the size of an input file right before a task is called however it works fine inside the task input attributes. Therefore all size() invocations are transferred to inside the tasks.
• Added flagger_end_to_end_no_variant_calling_no_ref_no_secphase.wdl

Related to HMM-Flagger (experimental):

• Added initial implementation for Negative Binomial emission model
• Added parameter alpha to relate the emission density of the current state to the previous observation (For Gaussian Model)

Flagger v0.3.0

• Gathered all miscellaneous files for Flagger in a single output
• Added a workflow for running Flagger end to end; including secphase, projecting necessary blocks, running preprocess, running flagger and calculating stats
• Updated Secphase to 0.3.0
• Added Flagger workflows without variant calling
• Added Flagger workflows without variant calling and projecting annotations
• Added an output text file for Flagger ngx plot
• Added a wdl for projecting blocks necessary for Flagger
• Added instructions for running Flagger locally

Flagger v0.2.0

Main changes in v0.2.0 release:

1. Because of the order inconsistencies that may exist between the bed files sorted by bedtools sort and the provided fasta file some flagger steps may not run properly. In this version Fasta files are being sorted with seqkit sort -nN before alignment in the WDL task developed for this aim, long_read_aligner.wdl. Because of the same reason fasta index tracks are sorted by cat ~{fai} | sort -V in cov2counts_by_window.wdl to obtain the bed file that covers the whole assembly.

2. The regions prone to coverage bias are removed prior to splitting assembly into windows (this line). So windows will no longer include those blocks and it can avoid affecting the parameters of the model fit for each window.

3. Added options to alignment wdls for aligning reads with methylation tags.
   ~{fastaOptions} to extract_reads.wdl, which should contain -y for methylation mode.
   ~{options} in long_read_aligner_scattered.wdl should contain -TMl,Mm.

4. As a new feature long_read_aligner_scattered.wdl can do the alignment with veritymap (by setting the parameter aligner to "veritymap"). The latest docker image have veritymap-v2.1.2-alpha installed.

5. deepvariant-v1.4 does not work with openvino so using openvino is added an option which is disabled by default.

6. Added a multi threaded version of project_blocks.py -> project_blocks_multi_thread.py

7. project_blocks_multi_thread.py is now able to project blocks with --eqx cigar format (with X and I)

8. Added FlaggerStats workflow. It receives the output of Flagger along with two main annotation bed files; presumably one for sex chromosome, one for centromeric regions and one for Seg Dups. It will output a tsv file containing flagger statistics like unreliablity percentage stratifed by the given annotations.

9. Fixed an issue for the windows (5-10Mb long) free of any alignment. The zero variance for those blocks crashes the pipeline. The variance is set to 1 for these cases to avoid division by zero.

10. Fixed some other minor bugs and typos in the workfows and scripts.

11. Added some initial scripts and workflows for HMM-Flagger which is still being tuned and experimented. HMM-Flagger uses a Hidden Markov Model to find the problematic regions based on read coverage.

Flagger v0.1

The first release of Flagger