Bam!

Changes from v0.43.1

BAM!

kallisto can now project pseudoalignments from transcripts down to genomic coordinates. This requires a GTF file corresponding to the transcriptome used to construct the index. The resulting BAM file is sorted by genomic coordinates and indexed.

• --pseudobam option works as before in transcript coordinates, but creates a single output pseudoalignments.bam in the output folder. This mode no longer writes SAM format to standard output, but writes the binary BAM file directly. Multithreaded --pseudobam works now
• --genomebam option writes pseudoalignments to the file pseudoalignments.bam in sorted genomic coordinates, requires a --gtf option and optionally a --chromosomes options set.

quant mode

Adds a --single-overhang option that does not discard reads where unobserved rest of fragment is predicted to lie outside a transcript. This is mainly useful for mapping 3' biased reads from single cell experiments.

JSON output

Adds QC information to run_info.json in the output folder

The added fields are

• n_pseudoaligned : number of fragments that could be pseudoaligned
• p_pseudoaligned : percentage of fragments that could be pseudoaligned
• n_unique : number of fragments that could be pseudoaligned to a unique target sequence
• p_unique : percentage of fragments that could be pseudoaligned to a unique target sequence

Fusion detection

Changes from v0.43.0

fusions

kallisto can now find reads which span potential fusion breakpoints. The quant mode adds a --fusion flag which identifies read pairs involved in fusions and writes output to fusion.txt, this file is then processed by pizzly for downstream analysis.

quant mode:

Switched to a uniform point for the EM algorithm that works better in highly ambiguous cases.

pseudobam fixes

Several fixes to the pseudobam output so that the resulting SAM/BAM file can be validated with picard.

Bug fixes

• updates kseq library, which would loop indefinitely on CRC corrupt gzipped files.
• warning when no reads pseudoalign and fixes crash (resulting output file will contain nan for tpm values).

Strand specific reads and new UMI mode for single cell transcriptomics

Changes from v0.42.5

quant mode:

Quantification can now be run in strand specific mode. The experiment can either be --fr-stranded, when the first read is on the forward strand or --rf-stranded
kallisto_mac-v0.43.0.tar.gz
when the first read is on the second strand.

pseudo mode:

• Improved multithreading when working with single cell data.
• A new UMI mode has been added which allows for efficient processing of UMI labelled single cell data.
• Sparse output for batch mode.

New pseudo mode for single-cell RNA-Seq analysis

pseudo mode:

A new mode kallisto pseudo has been implemented which allows reads to only be pseudoaligned. This mode is useful during single cell analysis as many different experiments (single cells) can be analyzed at the same time and their equivalence classes will be consistent. It will also output a matrix of equivalence class counts (as used in Fast and accurate single-cell RNA-Seq analysis by clustering of transcript-compatibility counts).

Bug fixes:

• Fixes a segfault that could occur when running quant in --bias mode
• Fixes a small error in allocation of memory
• Ensures that single-end reads have reasonable length mappings when mapping to short transcripts

Multithreaded pseudoalignment!

• multithreaded pseudoalignment
• bias parameters and fragment length distribution integrated into HDF5 (integration now exists in sleuth)
• bug fix when read maps to the end of a transcript

New features and effective length estimation

• We now use a conditional mean for effective length estimation (rather than the overall mean) based on the frag length distribution
• Add 'aux/num_processed' to H5
• Add GFA file option to 'inspect'
• Incorporate truncated gaussian in single-end datasets instead of only mean

v0.42.2.1 - Minor release with bugfixes to v0.42.2

A few bugfixes to v0.42.2 are included here:

• Off-by-one error fix in pseudobam output for reads mapping to reverse strand
• Rare segfault fix happening on Ubuntu 12.02
• -nan values would sometimes show up in abundances in edge case

It is recommended that users upgrade to this release.

Bias modeling, pseudobam output, and threaded bootstraps

This is version 0.42.2 of Kallisto.

Updates include

• index
  • the index has been updated to include more of the transcript information
  • indices should be slightly smaller than before
  • indices constructed with previous versions will not work with this version, rerun your index command
• quant
  • Bootstraps can be run in multithreaded mode, use option -t to specify number of threads
  • If only one read of a paired end maps, kallisto will check the transcript positions to discard reads that would go outside of transcript given the mean fragment length.
  • Pseudobam. Pseudoalignments can now be output in SAM format to standard output. For more details on the output see pseudobam
  • Sequence specific bias. kallisto can learn a model for sequence specific bias and correct the abundances accordingly.
  • TSV. All output text files have been changed to .tsv ending.

Minor update to include static binaries

This is version 0.42.1 of Kallisto. We recommend that users upgrade to this version.

Updates include

• index
  • now accepts multiple FASTA files and creates an index of all the targets
  • transcripts that end with polyA, runs of length 10 or greater have the polyA tail clipped
• quant
  • now accepts multiple FASTQ paired end files, or single-end files
  • --single flag to specify single-end reads
  • simplified jumping rules
  • uses soft-intersect, if one read has no mapping k-mers the other end is used for pseudoalignment

First release

First public release of kallisto.