One solution using SeqKit and shell commands join and awk.

1. Converting fasta files to tab-delimited format. Only FASTA ID instead of whole header is kept. Sorting is needed if you want join using shell command join later.

   $ cat df1.fa | seqkit sort | seqkit fx2tab -i > df1.fa.tsv
   $ cat df2.fa | seqkit sort | seqkit fx2tab -i > df2.fa.tsv
   
   $ cat df1.fa.tsv
   sp|O15304|SIVA_HUMAN    MPKRSCPFADVAPLQLKVRVSQRELSRGVCAERYSQEVFEKTKRLLFLGAQAYLDHVWDEGCAVVHLPESPKPGPTGAPRAARGQMLIGPDGRLIRSLGQASEADPSGVASIACSSCVRAVDGKAVCGQCERALCGQCVRTCWGCGSVACTLCGLVDCSDMYEKVLCTSCAMFET
   tr|A0A1B1L9R9|A0A1B1L9R9_BACTU  MNKQLFLASLKETQKSILSYACGAALYLWLLIWIFPSMVSAKGLNELIAAMPDSVKKIVGMESPIQNVMDFLAGEYYSLLFIIILTIFCVTVATHLIARHVDKGAMAYLLATPVSRVQIAITQATVLILGLLIIVSVTYVAGLVGAEWFLQDNNLNKELFLKINIVGGLIFLVVSAYSFFFSCICNDERKALSYSASLTILFFVLDMVGKLSDKLEWMKNLSLFTLFRPKEIAEGAYNIWPVSIGLIAGALCIFIVAIVVFKKRDLPL
   

2. Joining two TSV files. Using csvtk if IDs of records in the two files are not equal.

   # or:
   #   csvtk join -H -t df1.fa.tsv df2.fa.tsv 
   
   $ join df1.fa.tsv df2.fa.tsv        
   sp|O15304|SIVA_HUMAN    MPKRSCPFADVAPLQLKVRVSQRELSRGVCAERYSQEVFEKTKRLLFLGAQAYLDHVWDEGCAVVHLPESPKPGPTGAPRAARGQMLIGPDGRLIRSLGQASEADPSGVASIACSSCVRAVDGKAVCGQCERALCGQCVRTCWGCGSVACTLCGLVDCSDMYEKVLCTSCAMFET         IGPDGR
   tr|A0A1B1L9R9|A0A1B1L9R9_BACTU  MNKQLFLASLKETQKSILSYACGAALYLWLLIWIFPSMVSAKGLNELIAAMPDSVKKIVGMESPIQNVMDFLAGEYYSLLFIIILTIFCVTVATHLIARHVDKGAMAYLLATPVSRVQIAITQATVLILGLLIIVSVTYVAGLVGAEWFLQDNNLNKELFLKINIVGGLIFLVVSAYSFFFSCICNDERKALSYSASLTILFFVLDMVGKLSDKLEWMKNLSLFTLFRPKEIAEGAYNIWPVSIGLIAGALCIFIVAIVVFKKRDLPL            NKELFLKINIVGGLIFLVVSAYSFFFSCICNDERKALSYSASLTILFFVLDMVGKLSDKLEWM
   

3. Converting subsequence (3rd column) found in sequence (2nd column) to lowercase, and removing 3rd column.

   $ join df1.fa.tsv df2.fa.tsv \
       | awk '{ gsub($3, tolower($3)); print $1"\t"$2}'
   
   sp|O15304|SIVA_HUMAN    MPKRSCPFADVAPLQLKVRVSQRELSRGVCAERYSQEVFEKTKRLLFLGAQAYLDHVWDEGCAVVHLPESPKPGPTGAPRAARGQMLigpdgrLIRSLGQASEADPSGVASIACSSCVRAVDGKAVCGQCERALCGQCVRTCWGCGSVACTLCGLVDCSDMYEKVLCTSCAMFET
   tr|A0A1B1L9R9|A0A1B1L9R9_BACTU  MNKQLFLASLKETQKSILSYACGAALYLWLLIWIFPSMVSAKGLNELIAAMPDSVKKIVGMESPIQNVMDFLAGEYYSLLFIIILTIFCVTVATHLIARHVDKGAMAYLLATPVSRVQIAITQATVLILGLLIIVSVTYVAGLVGAEWFLQDNNLnkelflkinivggliflvvsaysfffscicnderkalsysasltilffvldmvgklsdklewmKNLSLFTLFRPKEIAEGAYNIWPVSIGLIAGALCIFIVAIVVFKKRDLPL
   

4. Converting TSV back to FASTA format.

   $ join df1.fa.tsv df2.fa.tsv \
       | awk '{ gsub($3, tolower($3)); print $1"\t"$2}' \
       | seqkit tab2fx \
       > df1.mask.fa
   
   $ cat df1.mask.fa
   >sp|O15304|SIVA_HUMAN
   MPKRSCPFADVAPLQLKVRVSQRELSRGVCAERYSQEVFEKTKRLLFLGAQAYLDHVWDE
   GCAVVHLPESPKPGPTGAPRAARGQMLigpdgrLIRSLGQASEADPSGVASIACSSCVRA
   VDGKAVCGQCERALCGQCVRTCWGCGSVACTLCGLVDCSDMYEKVLCTSCAMFET
   >tr|A0A1B1L9R9|A0A1B1L9R9_BACTU
   MNKQLFLASLKETQKSILSYACGAALYLWLLIWIFPSMVSAKGLNELIAAMPDSVKKIVG
   MESPIQNVMDFLAGEYYSLLFIIILTIFCVTVATHLIARHVDKGAMAYLLATPVSRVQIA
   ITQATVLILGLLIIVSVTYVAGLVGAEWFLQDNNLnkelflkinivggliflvvsaysff
   fscicnderkalsysasltilffvldmvgklsdklewmKNLSLFTLFRPKEIAEGAYNIW
   PVSIGLIAGALCIFIVAIVVFKKRDLPL
   

5. (Optional) If you want keep the full header.

   $ seqkit seq -n df1.fa | sed 's/ /\t/' > id2head.tsv
   
   $ cat df1.mask.fa | seqkit replace -k id2head.tsv -p '(.+)' -r '$1 {kv}' 
   >sp|O15304|SIVA_HUMAN Apoptosis regulatory protein Siva OS=Homo sapiens OX=9606 GN=SIVA1 PE=1 SV=2
   MPKRSCPFADVAPLQLKVRVSQRELSRGVCAERYSQEVFEKTKRLLFLGAQAYLDHVWDE
   GCAVVHLPESPKPGPTGAPRAARGQMLigpdgrLIRSLGQASEADPSGVASIACSSCVRA
   VDGKAVCGQCERALCGQCVRTCWGCGSVACTLCGLVDCSDMYEKVLCTSCAMFET
   >tr|A0A1B1L9R9|A0A1B1L9R9_BACTU ABC transporter permease OS=Bacillus thuringiensis OX=1428 GN=berB PE=4 SV=1
   MNKQLFLASLKETQKSILSYACGAALYLWLLIWIFPSMVSAKGLNELIAAMPDSVKKIVG
   MESPIQNVMDFLAGEYYSLLFIIILTIFCVTVATHLIARHVDKGAMAYLLATPVSRVQIA
   ITQATVLILGLLIIVSVTYVAGLVGAEWFLQDNNLnkelflkinivggliflvvsaysff
   fscicnderkalsysasltilffvldmvgklsdklewmKNLSLFTLFRPKEIAEGAYNIW
   PVSIGLIAGALCIFIVAIVVFKKRDLPL
   

Given sequences in in.fa, e.g.:

>sp|P14306|CPYI_YEAST Carboxypeptidase Y inhibitor OS=Saccharomyces cerevisiae (strain ATCC 204508 / S288c) OX=559292 GN=TFS1 PE=1 SV=2
MNQAIDFAQASIDSYKKHGILEDVIHDTSFQPSGILAVEYSSSAPVAMGNTLPTEKARSK
PQFQFTFNKQMQKSVPQANAYVPQDDDLFTLVMTDPDAPSKTDHKWSEFCHLVECDLKLL
NEATHETSGATEFFASEFNTKGSNTLIEYMGPAPPKGSGPHRYVFLLYKQPKGVDSSKFS
KIKDRPNWGYGTPATGVGKWAKENNLQLVASNFFYAETK

You could pipe them to the following script lc_pattern.py:

#!/usr/bin/env python

import sys
import re

try:
    pattern = sys.argv[1]
except ValueError as ve:
    sys.stderr.write("Error: Specify pattern\n")
    sys.exit(-1)

def process_record(header, needle):
    sequence = needle
    for idx in [m.start() for m in re.finditer(pattern, needle, re.IGNORECASE)]:
        offset = idx + len(pattern)
        sequence = sequence[:idx] + needle[idx:offset].lower() + sequence[offset + 1:]
    sys.stdout.write('%s\n%s\n' % (header, sequence))

sequence = ''
for line in sys.stdin:
    element = line.rstrip()
    if element.startswith('>'):
        header = element
        if len(sequence) > 0:
            process_record(header, sequence)
            sequence = ''
    else:
        sequence += element

# process last record
process_record(header, sequence)

For example:

$ python ./lc_pattern.py NTLIEYMGPAPPKGSG < in.fa > out.fa
$ more out.fa
>sp|P14306|CPYI_YEAST Carboxypeptidase Y inhibitor OS=Saccharomyces cerevisiae (strain ATCC 204508 / S288c) OX=559292 GN=TFS1 PE=1 SV=2
MNQAIDFAQASIDSYKKHGILEDVIHDTSFQPSGILAVEYSSSAPVAMGNTLPTEKARSKPQFQFTFNKQMQKSVPQANAYVPQD
DDLFTLVMTDPDAPSKTDHKWSEFCHLVECDLKLLNEATHETSGATEFFASEFNTKGSntlieymgpappkgsgHRYVFLLYKQP
KGVDSSKFSKIKDRPNWGYGTPATGVGKWAKENNLQLVASNFFYAETK