Extract

usage: extract [-h] [-m MODEL] [-r RADIUS] [-t THRESHOLD] [-s DOWN_SCALE]
               [-x UP_SCALE] [--num-workers NUM_WORKERS] [-j NUM_THREADS]
               [--batch-size BATCH_SIZE]
               [--assignment-radius ASSIGNMENT_RADIUS]
               [--min-radius MIN_RADIUS] [--max-radius MAX_RADIUS]
               [--step-radius STEP_RADIUS] [--targets TARGETS]
               [--only-validate] [-d DEVICE] [-o OUTPUT] [--per-micrograph]
               [--suffix SUFFIX] [--format {coord,csv,star,json,box}]
               [paths ...]

Positional Arguments

paths

paths to image files for processing, can also be streamed from stdin

Named Arguments

-m, --model

path to trained subimage classifier. uses the pretrained resnet16 model by default. if micrographs have already been segmented (transformed to log-likelihood ratio maps), then this should be set to “none” (default: resnet16)

Default: “resnet16”

-r, --radius

radius of the regions to extract

-t, --threshold

log-likelihood score threshold at which to terminate region extraction, -6 is p>=0.0025 (default: -6)

Default: -6

-s, --down-scale

DOWN-scale coordinates by this factor. output coordinates will be coord_out = (x/s)*coord. (default: 1)

Default: 1

-x, --up-scale

UP-scale coordinates by this factor. output coordinates will be coord_out = (x/s)*coord. (default: 1)

Default: 1

--num-workers

number of processes to use for extracting in parallel, 0 uses main process, -1 uses all CPUs (default: 0)

Default: 0

-j, --num-threads

number of threads for pytorch, 0 uses pytorch defaults, <0 uses all cores (default: 0)

Default: 0

--batch-size

batch size for scoring micrographs with model (default: 1)

Default: 1

--assignment-radius

maximum distance between prediction and labeled target allowed for considering them a match (default: same as extraction radius)

--min-radius

minimum radius for region extraction when tuning radius parameter (default: 5)

Default: 5

--max-radius

maximum radius for region extraction when tuning radius parameters (default: 100)

Default: 100

--step-radius

grid size when searching for optimal radius parameter (default: 5)

Default: 5

--targets

path to file specifying particle coordinates. used to find extraction radius that maximizes the AUPRC

--only-validate

flag indicating to only calculate validation metrics. does not report full prediction list

Default: False

-d, --device

which device to use, <0 corresponds to CPU

Default: 0

-o, --output

file path to write

--per-micrograph

write one particle file per micrograph at the location of the micrograph

Default: False

--suffix

optional suffix to add to particle file paths when using the –per-micrograph flag.

Default: “”

--format

Possible choices: coord, csv, star, json, box

file format of the OUTPUT files (default: coord)

Default: “coord”