Crate salmon_model 

salmon-model: statistical models used during quantification.

What these models are for

Sequencing is not a uniform sampler. A fragment’s chance of being observed depends on how long it is, what sequence sits at its ends, how GC-rich it is, and where in the transcript it starts. Left uncorrected, those effects are read as differences in abundance, which they are not. Each model here measures one such effect from the data itself and hands the quantifier a correction factor:

• fld — the fragment-length distribution, and automatic library-type detection (libdetect);
• seqbias — sequence-specific bias, from primer/ligation preferences at fragment ends;
• gcbias — fragment GC bias, largely from PCR amplification efficiency;
• posbias — positional bias along the transcript body, e.g. 3’ pileup from degraded RNA;
• bias — combines them into a corrected effective length per transcript.

Every model is observed vs expected: count what was actually seen, count what would have been seen under no bias, and take the ratio. That framing is why the dump files in dumps always come in obs/exp pairs.

Re-exports

pub use bias::build_expected_pos;

pub use bias::corrected_effective_length_full;

pub use bias::positional_factor;

pub use bias::positional_factor_into;

pub use bias::BiasInputs;

pub use fld::FragLengthSource;

pub use fld::ambig_frag_log_prob;

pub use fld::smoothed_effective_length;

pub use fld::DiscreteFld;

pub use fld::FragmentLengthDistribution;

pub use gcbias::build_expected_gc;

pub use gcbias::gc_corrected_effective_length;

pub use gcbias::gc_desc;

pub use gcbias::gc_prefix;

pub use gcbias::gc_ratio;

pub use gcbias::GcFragModel;

pub use gcbias::GcRank;

pub use gcbias::GcStore;

pub use gcbias::GcView;

pub use gcbias::GC_SAMP_STRIDE;

pub use libdetect::infer_format_from_counts;

pub use libdetect::LibraryTypeDetector;

pub use posbias::compute_length_quantiles;

pub use posbias::length_class_index;

pub use posbias::SimplePosBias;

pub use posbias::NUM_LENGTH_CLASSES;

pub use posbias::NUM_POS_BINS;

pub use seqbias::build_expected;

pub use seqbias::corrected_effective_length;

pub use seqbias::LogBiasTable;

pub use seqbias::SBModel;

Modules

bias

Unified bias-corrected effective length: composes sequence-specific, GC, and positional bias exactly as salmon’s updateEffectiveLengths does — a single conditional-FLD convolution whose per-fragment factor is the product of the enabled bias terms.

dumps

Aux-output dump helpers shared by reads-mode and alignment-mode quant.

fld

Fragment-length distribution.

gcbias

Fragment-GC bias model (GCFragModel).

libdetect

Automatic library-type detection.

posbias

Positional fragment bias (SimplePosBias) — a port of salmon’s src/model/SimplePosBias.cpp.

seqbias

Sequence-specific bias model (SBModel).

spline

Cubic spline interpolation — a faithful port of the vendored tk::spline (Tino Kluge) used by salmon’s SimplePosBias.

Constants

BIAS_WEIGHT_SCALE

Fixed-point scale for deterministic bias-model mass accumulation. Bias observed models sum per-fragment posterior masses (each in [0,1]) into bins; an f64 += is non-associative, so the accumulated model — and hence bias correction — varies with the worker-thread fragment partition (thread count). Accumulating round_down(mass * BIAS_WEIGHT_SCALE) as integers makes the sum associative (order/thread-count independent). 2^20 keeps the per-contribution resolution at ~1e-6 (ample for these coarse models) while a bin total (~num_fragments × 2^20) stays far below u64::MAX.

Functions

bias_mass_to_fp

Quantize a bias mass ([0,∞), typically a [0,1] posterior) to the fixed-point integer accumulator. Truncates (rounds toward zero) — cheaper than round() and the ≤1-ULP downward bias is negligible and cancels under the model’s normalization.