Sorting numpy array with np.lexsort containing float values

Question

I am writing a program to sort the names of amino acids depending on their energy value within a particular company.

I have extracted relevant data to the following numpy array.

And I tried this

In[37]: Data = np.array([
 ['ASN 205', -9.64164],
 ['LEU 206', -8.985774],
 ['ASN 207', -7.314434],
 ['PRO 208', -4.105338],
 ['ASN 209', -2.092342],
 ['GLY 210', -2.101412],
 ['LYS 211', -2.483852],
 ['ARG 212', -24.20364],
 ['SER 213', -1.181002],
 ['VAL 214', 0.057618]])
In[38]: ind3 = np.lexsort((Data[:,0],Data[:,1]))
In[39]: Result = Data[ind3]
In[40]: Result
Out[40]: 
array([['SER 213', '-1.181002'],
       ['ASN 209', '-2.092342'],
       ['GLY 210', '-2.101412'],
       ['LYS 211', '-2.483852'],
       ['ARG 212', '-24.20364'],
       ['PRO 208', '-4.105338'],
       ['ASN 207', '-7.314434'],
       ['LEU 206', '-8.985774'],
       ['ASN 205', '-9.64164'],
       ['VAL 214', '0.057618']], 
      dtype='|S9')

But the problem here is float values are arranged in a lexicographic way. I want it be be ordered according to their value means -24.20364 first then ...-2.483852.

How do I do this?

The regular numpy.sort says this :"order : list, optional When a is a structured array, this argument specifies which fields to compare first, second, and so on." — user707650
– user707650, Commented Jan 28, 2015 at 12:59
Your Data array has been immediately converted to all-string-array (dtype='|S9'). String is not a good format for floats. Think of a better data structure, such as a dict. — eumiro
– eumiro, Commented Jan 28, 2015 at 13:03
after rearranging my data I solved this problem using a crude method shown by my friend. How: TransposedData = numpy.transpose(Data) Result = Data[:,np.argsort(Data[1].astype(float))] Done — p.in4matics
– p.in4matics, Commented Jan 28, 2015 at 16:09

marc · Accepted Answer · 2015-01-28 13:04:17Z

2

explanation: np.array converts all the passed arguments to the biggest type fitting all arguments, i.e. your float get converted in line 1. You can instead use tuples with a specific data type as follows:

Data = np.array([
 ('ASN 205', -9.64164),
 ('LEU 206', -8.985774),
 ('ASN 207', -7.314434),
 ('PRO 208', -4.105338),
 ('ASN 209', -2.092342),
 ('GLY 210', -2.101412),
 ('LYS 211', -2.483852),
 ('ARG 212', -24.20364),
 ('SER 213', -1.181002),
 ('VAL 214', 0.057618)], dtype=[('f', '|S9'), ('g', float)])
ind3 = np.lexsort((Data['f'], Data['g']))
Result = Data[ind3]Out[8]:

output:

array([('ARG 212', -24.20364), ('ASN 205', -9.64164),
       ('LEU 206', -8.985774), ('ASN 207', -7.314434),
       ('PRO 208', -4.105338), ('LYS 211', -2.483852),
       ('GLY 210', -2.101412), ('ASN 209', -2.092342),
       ('SER 213', -1.181002), ('VAL 214', 0.057618)], 
      dtype=[('f', 'S9'), ('g', '<f8')])

answered Jan 28, 2015 at 13:04

marc

6,2431 gold badge31 silver badges33 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

p.in4matics Over a year ago

I am newbie, I reading this data from file, so all values are automatically converted to string. I got the tuple in this structure

(('HIE 203', '-1.889138'), ('TYR 204', '-2.148216'), ('ASN 205', '-9.64164'), ('LEU 206', '-8.985774'), ('ASN 207', '-7.314434'), ('PRO 208', '-4.105338'), ('ASN 209', '-2.092342'), ('GLY 210', '-2.101412'), ('LYS 211', '-2.483852'), ('ARG 212', '-24.20364'), ('SER 213', '-1.181002'), ('VAL 214', '0.057618'))

Now how to convert those string to float ? Sorry for this silly question.

marc Over a year ago

In case you are using numpy's loadtxt, you can just pass it the dtype=[('f', '|S9'), ('g', float)]) argument for the data type.

p.in4matics Over a year ago

Nice Idea. But the problem is from the file I am reading this data contains 18 columns, 2 of them string and rest floats. Is there any way to define data type for rest 16 columns in one go ? And one more thing, is that possible to skip reading some columns. ? BTW I am using numpy.genfromtext.

Collectives™ on Stack Overflow

Sorting numpy array with np.lexsort containing float values

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related