Default parameters¶
The parameters and options decribed below are all optional. All parameters and options have default values that do not normally need to be modified.
Important
Rp-Bp parameters are changed via the configuration file, but options for programs such as Flexbar and STAR are handled via command line arguments. You do not need to include Rp-Bp parameters in the configuration file, unless you wish to change their values.
Flexbar and STAR options¶
Default options are overridden via command line using --flexbar-options or --star-options. Currently, no options can be passed to Bowtie2.
Flexbar¶
max-uncalledDefault: 1.pre-trim-leftDefault: 0.qtrim-formatDefault: sanger.qtrimDefault: TAIL.qtrim-thresholdDefault: 10.zip-outputDefault: GZ.
STAR¶
readFilesCommandDefault: zcat (gunzip -c for macOS).limitBAMsortRAMDefault: 0 (set to--memat run-time).alignIntronMinDefault: 20.alignIntronMaxDefault: 100000.outFilterMismatchNmaxDefault: 1.outFilterMismatchNoverLmaxDefault: 0.04.outFilterTypeDefault: BySJout.outFilterIntronMotifsDefault: RemoveNoncanonicalUnannotated.outSAMattributesDefault: AS NH HI nM MD.outSAMtypeDefault: BAM SortedByCoordinate.sjdbOverhangDefault: 33.seedSearchStartLmaxOverLreadDefault: 0.5.winAnchorMultimapNmaxDefault: 100.
Rp-Bp parameters¶
keep_riboseq_multimappersIfTruein the configuration file, then multimapping riboseq reads will not be removed. They will be treated as “normal” reads in every place they map, i.e. the weight of the read will not be distributed fractionally.models_baseThe path to the compiled models, if installed in a different location. The models are included with the source distribution and compiled as part of the installation. Do not change this, unless you know what you are doing!
Metagene and periodicity estimation parameters¶
metagene_start_upstreamThe number of bases upstream of the translation initiation site to begin constructing the metagene profile. Default: 300.metagene_start_downstreamThe number of bases downstream of the translation initiation site to end the metagene profile. Default: 300.metagene_end_upstreamThe number of bases upstream of the translation termination site to begin constructing the metagene profile. Default: 300.metagene_end_downstreamThe number of bases downstream of the translation termination site to end the metagene profile. Default: 300.periodic_offset_startThe position, relative to the translation initiation site, to begin calculating periodicity Bayes factors. Default: -20 (inclusive).periodic_offset_endThe position, relative to the translation initiation site, to stop calculating periodicity Bayes factors. Default: 0 (inclusive).metagene_profile_lengthThe length of the profile to use in the models.metagene_profile_length+periodic_offset_endmust be consistent with the length of the extracted metagene profile. Default: 21.metagene_iterationsThe number of iterations to use for each chain in the MCMC sampling. Default: 500 (includes warmup).min_metagene_profile_countRead lengths with fewer than this number of reads will not be used. Default: 1000.min_metagene_bf_meanIfmax_metagene_bf_varandmin_metagene_bf_likelihoodare None (null in YAML), this is taken as a hard threshold on the estimated Bayes factor mean. Default: 5.max_metagene_bf_varA hard threshold on the estimated Bayes factor variance. Default: None.min_metagene_bf_likelihoodA threshold on the likelihood of periodicity. Default: 0.5.
Note
A profile is periodic if [P(bf > min_metagene_bf_mean)] > min_metagene_bf_likelihood. By default, we do not filter on the variance. If given, then both filters are applied and the result is the intersection.
Fixed lengths and offsets¶
use_fixed_lengthsIfTruein the configuration file, fixed values given bylengthsandoffsetsare used (no periodicity estimation).lengthsA list of read lengths to use for creating the profiles if theuse_fixed_lengthsoption isTrue. Presumably, these are lengths that have periodic metagene profiles.offsetsThe P-site offset to use for each read length specifed bylengthsif theuse_fixed_lengthsoption isTrue. The number of offsets must match the number of lengths, and they are assumed to match. For examplelengths: [26, 29] withoffsets: [9, 12] means only reads of lengths 26 bp and 29 bp are used to create the profiles. The 26 bp reads will be shifted by 9 bp in the 5’ direction, while reads of length 29 bp will be shifted by 12 bp.
Smoothing parameters¶
smoothing_fractionThe fraction of the data used when estimating each y-value for LOWESS. Default: 0.2.smoothing_reweighting_iterationsThe number of residual-based reweightings to perform for LOWESS. See the statsmodels documentation. Default: 0.
Translation prediction parameters¶
orf_min_length_preORFs with length <orf_min_length_pre(nucleotides) are not processed. Default: 0 (ignore option).orf_max_length_preORFs with length >orf_max_length_pre(nucleotides) are not processed. Default: 0 (ignore option).orf_min_lengthOnly ORFs with length >orf_min_length(nucleotides) are kept in the final set. Default: 8.orf_min_profile_count_preORF with profile sum <orf_min_profile_count_preare not processed. Default: 5.orf_min_profile_countOnly ORFs with profile sum >orf_min_profile_countare kept in the final set. Default: None.translation_iterationsThe number of iterations to use for each chain in the MCMC sampling. Default: 500 (includes warmup).min_bf_meanIfmax_bf_varandmin_bf_likelihoodare None (null in YAML), this is taken as a hard threshold on the estimated Bayes factor mean. Default: 5.max_bf_varA hard threshold on the estimated Bayes factor variance. Default: None.min_bf_likelihoodA threshold on the likelihood to select an ORF as translated. Default: 0.5.chisq_alphaFor the chi-square test, this value is first Bonferroni corrected based on the number of ORFs which pass the smoothing filters. It is then used as the significance threshold to select translated ORFs. Default: 0.01.
Note
A Ribo-seq ORF is translated if [P(bf > min_bf_mean)] > min_bf_likelihood. By default, we do not filter on the variance. If given, then both filters are applied and the result is the intersection.
Attention
Chi-square values are reported, but they are not used for prediction, unless the chi_square_only flag is present in the configuration file, in which case the translation models are not fit to the data, and the posterior distributions are not estimated. This is mostly kept for historical reasons, and may eventually be removed.