Daniel Fürth tractatus

## tensorflow-source-windows.md

      
              1 file
            
          
              0 forks
            
          
              0 comments
            
          
              0 stars
            
          
                tractatus
                / tensorflow-source-windows.md
            
            
              Last active
              June 18, 2021 23:15
            
              
                Speed-up by compiling TensorFlow from source?
              
          
    Why compiling from source?

The most straigtforward way install TensorFlow is to work in a virtual Python environment and simply to use either the TensorFlow official packages in pip or a wheel (.whl) distribution.
Problems arise at some point when a lot of heavy image processing needs to be done.
Since prediction/inference primarly involves the CPU we might correctly ask ourselves:
Instead of precompiled binaries, that are not optimized for our set of CPU architecture extensions,
what would be the speed gain if we would compile TensorFlow from source?

  
## U13369.1.fasta
>RNA5-8SN5 URS0000005270_9606 Homo sapiens RNA, 5.8S ribosomal N1 (RNA5-8SN 1 to 3, RNA5-8SN5)
CGACTCTTAGCGGTGGATCACTCGGCTCGTGCGTCGATGAAGAACGCAGCTAGCTGCGAG
AATTAATGTGAATTGCAGGACACATTGATCATCGACACTTCGAACGCACTTGCGGCCCCG
GGTTCCTCCCGGGGCTACGCCTGTCTGAGCGTCGCTT
>18S URS000031689A_9606 Homo sapiens small subunit ribosomal RNA (18S)
TACCTGGTTGATCCTGCCAGTAGCATATGCTTGTCTCAAAGATTAAGCCATGCATGTCTA
AGTACGCACGGCCGGTACAGTGAAACTGCGAATGGCTCATTAAATCAGTTATGGTTCCTT
TGGTCGCTCGCTCCTCTCCTACTTGGATAACTGTGGTAATTCTAGAGCTAATACATGCCG
ACGGGCGCTGACCCCCTTCGCGGGGGGGATGCGTGCATTTATCAGATCAAAACCAACCCG
GTCAGCCCCTCTCCGGCCCCGGCCGGGGGGCGGGCGCCGGCGGCTTTGGTGACTCTAGAT

## gist:04557e07725674b81cebc51b1b9a0893
#load data
data<-read.table('/Users/danielfurth/Downloads/fisseq_output/results.tsv', sep='\t', header=TRUE )
#size select rolonies
data<-data[data$size>4,]
#extract single cycle info
intensity<-lapply(data$intensity,function(x)strsplit(x, ","))
read<-lapply(data$read,function(x)strsplit(x, ","))
comp<-lapply(data$component,function(x)strsplit(x, ","))
qual<-lapply(data$b_quality_pbm,function(x)strsplit(x, ","))

## index.html
<!doctype html>
<html lang="en">
  <head>
    <meta charset="utf-8">
    <meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
    <meta name="description" content=".">
    <meta name="author" content="Daniel Furth">
    <link rel="icon" href="favicon.ico">
    <title>Segment</title>

## readBioanalyzer.R
library(drc) #used for fitting Weibull retardation model to get Time vs. Size

clean.up<-function(data){
  data<-data[-(nrow(data):(nrow(data)-1)),]
  data$Value<-as.numeric(as.vector(data$Value))
  data$Time<-as.numeric(as.vector(data$Time))
  return(data)
}

find_peaks <- function (data, nt = c(25, 200, 500, 1000, 2000, 4000, 6000),  m = 50, k = 2){

## Stitch WholeBrain Perkin-Elmer
#' Sticthing on a Perkin-Elmer instrument
#'
#' This function allows you to stitch image tiles from Perkin Elmer Harmony software contained in a single folder.
#' @param folder a path to a folder containing image tiles in TIFF format and XML file for metadata.
#' @param digits number of significant digits to round up the position X,Y reported by Harmony XML metadata (this is used for matching).
#' @return a character vector with images that have been stitched.
#' @keywords perkin-elmer
#' @export
#' @examples
#' stitch.perkin.elmer('./Images')

## Stationary Wavelet Experimental
imgSWT <- function(input, filter=NULL, scales=6, cell.bodies = 3, processes = 5, family='db2', sigma=10, processLength=12, alim=c(200, 500), pch=21, bg="white", cex=2, lwd=2, illustrator=F, output='waveletoutput') {
    file <- as.character(input)[1]
    family <- as.character(family)[1]
    outputfile <- as.character(output)[1]
    scales<-as.integer(scales)
    cellBodies<-as.integer(cell.bodies)
    processes <-  as.integer(processes)
    sigmaR<-as.integer(sigma)
    processLength<-as.integer(processLength)
    areaSize<-alim

## ameet.R
library(wholebrain)
#set folder path
folder<-'/Users/ameetrahane/Lab/Widlbrecht_Lab/brain-tissue-analysis/sectioned_test_1'
setwd(folder)
subfolders <- list.dirs(path = ".", full.names = TRUE, recursive = TRUE)
data <- data.frame()

#map to atlas coordinates
#set cutting thickness in millimeters
distance.between.sections <- 0.1

## edit_seg_in_script.R
#just change the numbers here.
# alim = area limits of the soma, lower and upper
# threshold.range = where in intesnity to look for neurons, lower and upper.
# eccentricity = the roundness of neurons, lower means rounder.
# Max = the maximum intensity to do 8 bit rendering on (just for display purposes).
# Min = the minimum intensity to do 8 bit rendering on (just for display purposes).
# brain.threshold = the intensity where you find the autofluorescent outline of the brain section.
# resize = resizing parameter to match atlas and your brain section.
# blur = blurring (higher = more) to apply to the contour of the brain if you have damaged tissue etc.
# downsample = how much to downsample the image to reduce image processing speed.

## readroi.R
##    readroi.R - Read ImageJ files in to R
##    Copyright (C) 2011 David C. Sterratt <david.c.sterratt@ed.ac.uk>

##    This program is free software: you can redistribute it and/or modify
##    it under the terms of the GNU General Public License as published by
##    the Free Software Foundation, either version 3 of the License, or
##    (at your option) any later version.

##    This program is distributed in the hope that it will be useful,
##    but WITHOUT ANY WARRANTY; without even the implied warranty of
	>RNA5-8SN5 URS0000005270_9606 Homo sapiens RNA, 5.8S ribosomal N1 (RNA5-8SN 1 to 3, RNA5-8SN5)
	CGACTCTTAGCGGTGGATCACTCGGCTCGTGCGTCGATGAAGAACGCAGCTAGCTGCGAG
	AATTAATGTGAATTGCAGGACACATTGATCATCGACACTTCGAACGCACTTGCGGCCCCG
	GGTTCCTCCCGGGGCTACGCCTGTCTGAGCGTCGCTT
	>18S URS000031689A_9606 Homo sapiens small subunit ribosomal RNA (18S)
	TACCTGGTTGATCCTGCCAGTAGCATATGCTTGTCTCAAAGATTAAGCCATGCATGTCTA
	AGTACGCACGGCCGGTACAGTGAAACTGCGAATGGCTCATTAAATCAGTTATGGTTCCTT
	TGGTCGCTCGCTCCTCTCCTACTTGGATAACTGTGGTAATTCTAGAGCTAATACATGCCG
	ACGGGCGCTGACCCCCTTCGCGGGGGGGATGCGTGCATTTATCAGATCAAAACCAACCCG
	GTCAGCCCCTCTCCGGCCCCGGCCGGGGGGCGGGCGCCGGCGGCTTTGGTGACTCTAGAT
	#load data
	data<-read.table('/Users/danielfurth/Downloads/fisseq_output/results.tsv', sep='\t', header=TRUE )
	#size select rolonies
	data<-data[data$size>4,]
	#extract single cycle info
	intensity<-lapply(data$intensity,function(x)strsplit(x, ","))
	read<-lapply(data$read,function(x)strsplit(x, ","))
	comp<-lapply(data$component,function(x)strsplit(x, ","))
	qual<-lapply(data$b_quality_pbm,function(x)strsplit(x, ","))
	<!doctype html>
	<html lang="en">
	<head>
	<meta charset="utf-8">
	<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
	<meta name="description" content=".">
	<meta name="author" content="Daniel Furth">
	<link rel="icon" href="favicon.ico">
	<title>Segment</title>
	library(drc) #used for fitting Weibull retardation model to get Time vs. Size

	clean.up<-function(data){
	data<-data[-(nrow(data):(nrow(data)-1)),]
	data$Value<-as.numeric(as.vector(data$Value))
	data$Time<-as.numeric(as.vector(data$Time))
	return(data)
	}

	find_peaks <- function (data, nt = c(25, 200, 500, 1000, 2000, 4000, 6000), m = 50, k = 2){
	#' Sticthing on a Perkin-Elmer instrument
	#'
	#' This function allows you to stitch image tiles from Perkin Elmer Harmony software contained in a single folder.
	#' @param folder a path to a folder containing image tiles in TIFF format and XML file for metadata.
	#' @param digits number of significant digits to round up the position X,Y reported by Harmony XML metadata (this is used for matching).
	#' @return a character vector with images that have been stitched.
	#' @keywords perkin-elmer
	#' @export
	#' @examples
	#' stitch.perkin.elmer('./Images')
	imgSWT <- function(input, filter=NULL, scales=6, cell.bodies = 3, processes = 5, family='db2', sigma=10, processLength=12, alim=c(200, 500), pch=21, bg="white", cex=2, lwd=2, illustrator=F, output='waveletoutput') {
	file <- as.character(input)[1]
	family <- as.character(family)[1]
	outputfile <- as.character(output)[1]
	scales<-as.integer(scales)
	cellBodies<-as.integer(cell.bodies)
	processes <- as.integer(processes)
	sigmaR<-as.integer(sigma)
	processLength<-as.integer(processLength)
	areaSize<-alim
	library(wholebrain)
	#set folder path
	folder<-'/Users/ameetrahane/Lab/Widlbrecht_Lab/brain-tissue-analysis/sectioned_test_1'
	setwd(folder)
	subfolders <- list.dirs(path = ".", full.names = TRUE, recursive = TRUE)
	data <- data.frame()

	#map to atlas coordinates
	#set cutting thickness in millimeters
	distance.between.sections <- 0.1
	#just change the numbers here.
	# alim = area limits of the soma, lower and upper
	# threshold.range = where in intesnity to look for neurons, lower and upper.
	# eccentricity = the roundness of neurons, lower means rounder.
	# Max = the maximum intensity to do 8 bit rendering on (just for display purposes).
	# Min = the minimum intensity to do 8 bit rendering on (just for display purposes).
	# brain.threshold = the intensity where you find the autofluorescent outline of the brain section.
	# resize = resizing parameter to match atlas and your brain section.
	# blur = blurring (higher = more) to apply to the contour of the brain if you have damaged tissue etc.
	# downsample = how much to downsample the image to reduce image processing speed.
	## readroi.R - Read ImageJ files in to R
	## Copyright (C) 2011 David C. Sterratt <david.c.sterratt@ed.ac.uk>

	## This program is free software: you can redistribute it and/or modify
	## it under the terms of the GNU General Public License as published by
	## the Free Software Foundation, either version 3 of the License, or
	## (at your option) any later version.

	## This program is distributed in the hope that it will be useful,
	## but WITHOUT ANY WARRANTY; without even the implied warranty of