Source code for acqpack.utils

import csv
import json

import numpy as np
import pandas as pd

[docs]def read_delim(filepath): """ Reads delimited file (auto-detects delimiter + header). Returns list. :param filepath: (str) location of delimited file :return: (list) list of records w/o header """ f = open(filepath, 'r') dialect = csv.Sniffer().sniff( has_header = csv.Sniffer().has_header( reader = csv.reader(f, dialect) if has_header: ret = [line for line in reader] return ret
[docs]def read_delim_pd(filepath): """ Reads delimited file (auto-detects delimiter + header). Returns pandas DataFrame. :param filepath: (str) location of delimited file :return: (DataFrame) """ f = open(filepath) has_header = None if csv.Sniffer().has_header( has_header = 0 return pd.read_csv(f, header=has_header, sep=None, engine='python')
[docs]def lookup(table, lookup_cols, lookup_vals, output_cols=None, output_recs=None): """ Looks up records where lookup_cols == lookup_vals. Optionally returns only specified output_cols and/or specified output_recs. :param table: (DataFrame) the pandas DataFrame to use as a lookup table :param lookup_cols: (str | list) :param lookup_vals: (val | list) :param output_cols: :param output_recs: :return: """ if type(lookup_cols) == str: lookup_cols = [lookup_cols] lookup_vals = [lookup_vals] temp_df = pd.DataFrame(data=lookup_vals, columns=lookup_cols, copy=False) output = table.merge(temp_df, copy=False) if output_cols is not None: if type(output_cols) == str: output_cols = [output_cols] output = output[output_cols] if output_recs is not None: output = output.iloc[output_recs] return output
[docs]def generate_position_table(num_rc, space_rc, offset=(0.0,0.0,0.0), to_clipboard=False): """ Generates a position table for a plate. Assumes that 'x' and 'c' are aligned and that 'y' and 'r' are aligned. These axes can be reflected by negating the corresponding 'space_rc'; translations can be applied via 'offset'. All entries are indexed by 'n' (newspaper order) and 's' (serpentine order). Other columns may be added as needed, but Autosampler.goto() requires 'x', 'y', and 'z' to function properly. :param num_rc: (tup) number of rows and columns (num_rows, num_cols) :param space_rc: (tup) spacing for rows and columns [mm] (spacing_rows, spacing_cols) :param offset: (tup) 3-tuple of floats to be added to x,y,z [mm] :param to_clipboard: (bool) whether to copy the position_table to the OS clipboard :return: (DataFrame) """ # TODO: instead of offset, full affine option? can use negative space rc to reflect, # but can't remap x -> y temp = list() headers = ['n', 's', 'r', 'c', 'name', 'x', 'y', 'z'] for r in range(num_rc[0]): for c in range(num_rc[1]): n = c + r * num_rc[1] s = ((r + 1) % 2) * (c + r * num_rc[1]) + (r % 2) * ((r + 1) * num_rc[1] - (c + 1)) name = chr(64 + r + 1) + '{:02d}'.format(c + 1) x = float(c * space_rc[1] + offset[0]) y = float(r * space_rc[0] + offset[1]) z = float(offset[2]) temp.append([n, s, r, c, name, x, y, z]) position_table = pd.DataFrame(temp, columns=headers) if to_clipboard: position_table.to_clipboard(index=False) return position_table
[docs]def spacing(num_rc, p1, p2): r, c = map(float, num_rc) return tuple(abs(np.nan_to_num(np.subtract(p2, p1) / (c - 1, r - 1))))
[docs]def load_mm_positionlist(filepath): """ Takes a MicroManager position list and converts it to a pandas DataFrame. Will load z-coordinates if available. :param filepath: (str) :return: (DataFrame) position list with headers = "r, c, name, x, y, [z]" """ with open(filepath) as f: data = json.load(f) df_rcn =, ['POSITIONS'])[['GRID_ROW', 'GRID_COL', 'LABEL']] df_pos =, ['POSITIONS', 'DEVICES'])[['DEVICE', 'X', 'Y']] df_xy = df_pos.query("DEVICE=='XYStage'")[['X','Y']].reset_index(drop=True) df = pd.concat([df_rcn,df_xy], axis=1) # check for z-axis ds_z = df_pos.query("DEVICE=='ZStage'")['X'].reset_index(drop=True) if len(ds_z)>0: df['z'] = ds_z rename = {'GRID_ROW': 'r', 'GRID_COL': 'c', 'LABEL': 'name', 'X': 'x', 'Y': 'y'} df.rename(columns=rename, inplace=True) return df
[docs]def generate_grid(c0, c1, l_img, p): """ Based on two points, creates a 2D-acquisition grid similar to what MicroManager would produce. :param c0: (arr) first point; numpy 1d array of len 2 :param c1: (arr) second point; numpy 1d array of len 2 :param l_img: (float) :param p: (float) desired percent overlap :return: (DataFrame) position_list in the same format as load_mm_positionlist """ # TODO: does generate_grid subsume generate_position_table? # n -> number of stage positions on an axis n = 1 + np.ceil(np.abs(c1 - c0) / ((1 - p) * l_img)) # ct,ct n = n.astype('int') # l_acq = total_movement + l_img # l_acq = l_img * (n - n*p + p) # um,um sign = np.sign(c1 - c0) # could also use cartesian product (itertools.product OR np.mgrid, stack) # position_list = pd.DataFrame(columns=['r', 'c', 'name', 'x', 'y'], ) for j in xrange(n[1]): # iter y y = sign[1] * j * l_img * (1 - p) + c0[1] for i in xrange(n[0]) if not (j % 2) else reversed(xrange(n[0])): # iter x (serp) x = sign[0] * i * l_img * (1 - p) + c0[0] r = j c = i name = '1-Pos_{:03}_{:03}'.format(c, r) position_list.loc[len(position_list)] = [r, c, name, x, y] position_list[['r', 'c']] = position_list[['r', 'c']].astype(int) return position_list