visualization.md

October 6, 2015 ยท View on GitHub

Basset

Deep convolutional neural networks for DNA sequence analysis.

Visualization

###### basset_motifs.py

Collect statistics and make plots to explore the first convolution layer of the given model using the given sequences.

ArgumentTypeDescription
model_fileModelTrained model
test_hdf5_fileHDF5Test data
OptionsVariableDescription
-dmodel_hdf5_filePre-computed model output as HDF5
-oout_dirOutput directory
-mmeme_dbMEME database used to annotate motifs
-ssampleSample sequences from the test set
-ttrim_filtersTrim uninformative positions off the filter ends

###### basset_motifs_infl.py

Collect statistics and make plots to explore the first convolution layer of the given model using the given sequences.

ArgumentTypeDescription
model_fileModelTrained model
test_hdf5_fileHDF5Test data
OptionsVariableDescription
-bbatch_sizeBatch size (affects memory usage) [Default: 1000]
-dmodel_hdf5_filePre-computed model output as HDF5
-iinformative_onlyPlot informative filters only
-mmotifs_fileMotifs table file output by basset_motifs.py
-nnorm_targetsUse the norm of the target influences as the primary influence measure
-oout_dirOutput directory
--subsetsubset_fileSubset targets to those in this file
-ssampleSample sequences from the test set
-ttargets_fileFile specifying target indexes and labels
--widthheat_widthHeatmaps width [Default: 10]
--heightheat_heightHeatmaps height [Default: 10]
--fontheat_fontHeatmaps font size [Default: 0.4]

###### basset_sat.py

Perform an in silico saturated mutagenesis of the given test sequences using the given model.

ArgumentTypeDescription
model_fileModelTrained model
input_fileFASTA or HDF5Test data
OptionsVariableDescription
-ainput_activity_fileOptional activitiy table matching an input FASTA file
-dmodel_hdf5_filePre-computed model output as HDF5
-mmin_limitMinimum heat map limit [Default: 0.1]
-ncenter_ntCenter nt to mutate and plot in the heat map [Default: 200]
-oout_dirOutput directory
-ssampleSample sequences from the test set
-ttargetsComma-separated list of target indexes to plot (or -1 for all)

###### basset_sat_vcf.py

Perform an in silico saturated mutagenesis of the regions surrounding a list of SNPs given in VCF format using the given model.

ArgumentTypeDescription
model_fileModelTrained model
vcf_fileVCFSNPs
OptionsVariableDescription
-dmodel_hdf5_filePre-computed model output as HDF5
-fgenome_fastaGenome FASTA from which sequences will be drawn
-lseq_lenSequence length provided to the model
-mmin_limitMinimum heat map limit [Default: 0.1]
-ncenter_ntNt around the SNP to mutate and plot in the heat map [Default: 200]
-oout_dirOutput directory
-ttargetsComma-separated list of target indexes to plot (or -1 for all)

###### basset_sad.py

Compute SNP Accessibility Difference scores for SNPs in a VCF file using the given model.

ArgumentTypeDescription
model_fileModelTrained model
vcf_fileVCFSNPs
OptionsVariableDescription
-dmodel_hdf5_filePre-computed model output as HDF5
-fgenome_fastaGenome FASTA from which sequences will be drawn
-iindex_snpSNPs are labeled with their index SNP in column 6
-lseq_lenSequence length provided to the model
-mmin_limitMinimum heat map limit [Default: 0.1]
-ncenter_ntNt around the SNP to mutate and plot in the heat map [Default: 200]
-oout_dirOutput directory
-sscoreSNPs are labeld with scores as column 7